One thing is to train the model, but another is to serve it. In between those two phases, model serialization and deserialization occurs. In simpler words it’s just model saving and loading. It’s important because different methods result in different:

1. Inference speed
2. Model size
3. Python environment size

If you are curious what are the ways to serialize model in PyTorch and how they compare, checkout my new post on Appsilon blog.

Finally the recording of the talk StyleGAN w Twoich rękach from SFI 2022 is available online! 🎉

SFINKS is a small conference on complex systems and interdisciplinary physics. I wanted to present something there since the first edition in 2016, but I never stood a chance. Every year there was something up, like an international exchange or a lot of tough exams. Finally, this year I was able to fullfil my dream as I presented a lightning talk O strachu przed liczeniem na papierze (eng. About the fear of calculating on paper).

I’d like to inform that the course has finished and the total of 56 people received a graduation certificate (which required passing homeworks for over 70%), congratulations! 🎉 It was a great experience and I’m happy that I was able to teach so many young students!

I’m happy to announce that I’ll create and teach the course Python LevelUP: Data Science by Appsilon in collaboration with DaftAcademy, part of DaftCode.

This will be a five-week course on the basics of Data Science in python. The course materials will be available on GitHub for free. Apart from teaching the classic stuff, I’ll try to show some of the newest and coolest libraries out there. Even if you already know most of this stuff, I really recommend to take a closer look on the unknown packages.

Some time ago, I’ve worked on an oil sac (energy reservoir) detection in copepods (kind of zooplankton) for Université Laval. Apart from presenting the results on the ML4Plankton conference I’ve published a detailed blog post on the Appsilon blog that I strongly encourage you to read!