I'm starting my first day by picking up the fast ai "Practical Deep Learning for Coders" course. I have experience in web development and I know the basics of python so it seems like a good place to begin.
So far I enjoy the videos and the top-down approach.
I've logged into my dusty Kaggle account and started experimenting with lesson 1 material.
Unfortunately, the first notebook includes a deprecated method from the duck duck go library for image search. There are some open pull requests on the course repo, but nobody seems to be maintaining it.
Eventually, I found the answer in the comments section under the original notebook. At the time of writing, I got it to work by replacing the original ddg_images() implementation with a method directly from the fastbook package.
... and I could finally see the happy parrots :)
Insert the "It ain't much, but it's honest work" meme here...
What I've learned today:
- how to setup and use a Jupyter Notebook / Kaggle cloud environment
- overview of the fastai library (what is DataBlock, Learner, etc.)
- what are some common use case examples for machine learning such as computer vision, segmentation, tabular data, collaborative filtering (recommendation systems) etc.
Time commitment:
The video was around 1 hour long (at 1.25x speed), and playing with the notebook took up another hour.
Total: 2 hours
Hopefully, I can complete the first lesson tomorrow. Wish me luck!