Right before the arrival of the first LLMs, I co-authored a book on Model Risk from a machine learning perspective. Rather than taking a (purely) financial view on the topic, we wanted to break open the term Model Risk to cover all predictive modelling failure cases that lead to harm …
I have a nasty habbit of starting quick new Python projects somewhere in my chronologically organized Google Drive, uv init-ing them, and start adding packages.
By default, this means every project gets a .venv directory alongside its pyproject.toml. That works perfectly on a single machine. It also works …
Yesterday (6 Jan), someone showed off a small clever puzzle game on Hacker News, called enclose.horse.
An example puzzle looks like this:
The game is played on a grid of grass or water tiles. The goal is to place walls so that the horse cannot escape (i.e. exit …
The last time I did this was in 2017, but I’m in some mood to do it once again. With Summer Games Fest, Next Fest, the PC Gaming Show, the Future Games Show, Fan Fest, and so, so many more shows, even a hardcore fan of gaming would be …
At the end of 2023, certain events in my life led to a realization that I was spending way too little time on reading in the years past, even though I have always been an avid reader in my youth and adult life. Facing some “weltschmerz” and not finding enjoyment …
If you’re at all interested in procedural generation, digital art or game development, it’s very likely that you’ve heard about the wave function collapse (WFC) algorithm before. It’s incredible popular, has been ported to virtually every language and engine, and used in several games.
A friend of mine recently asked whether there was an easy way to introduce new levels (i.e. items) to an embedding layer in Keras and only training the embedding layer for those new levels, i.e. keeping the weights for the existing indices fixed.
Note that this is not …
