aburla

In this article I tell the story of how I came to codify my DL research setup and dive into the technical decisions that shaped it. I hope many of you will find it useful, whether you're looking for alternatives to Google Colab, having your own GPU workstation, or even looking for ways to make the most out of cloud GPUs.

Any tools you would recommend for the scenario where you'd use monorepos?

In this article I tried to summarize my personal thought process when deploying machine learning models. It's not quite a tutorial but can give you some good directions when faced with the problem of deciding how to serve a machine learning model.

Any feedback or questions are greatly appreciated. Hope you'll find it useful!

A few months ago I wrote an article about some considerations of how to perform error analysis for a trained machine learning model. It also proposes a “framework”/progression ladder for what methods to implement first.

I’d be happy if some of you find it useful, or maybe has some comments/questions.

I find having a simple baseline, and ideally a human baseline, extra important for a real world project. I had once the experience of working on a complicated project without any human baseline, even when requested. It was like walking through a dark forest without any lights. A lot of bruises.

Basically I assume a good AutoML solution, given time and enough compute, can come up with a very good model, much better than I can do in a few weeks, assuming no feature engineering. This way I can focus more on what could be missing and I don't really have to think "what if I used more estimators in that random forest" et al.

There's no formal "estimation".