This weekends project, generate extruded 3d meshes from rich text for use in the Bevy Game Engine

https://github.com/terhechte/bevy-mesh-text-3d

I asked qwen 0.5B, a 800MB model the if it knows the current month. This was possible by giving it a short "system prompt" of "context information" (city, datetime, weather, etc).

It has no problem answering this.

Just how bad is the model Apple Intelligence is using behind the scenes. (cc @gruber )

I *just* finished a benchmark study to evaluate the impact of programming language choice on the quality of various Language Learning Models (LLMs). The findings were quite intriguing:
For instance, there was a 20% quality difference between Python and Rust when tested with the o3-mini model, with even more pronounced differences observed with other LLMs.

https://ben.terhech.de/posts/2025-01-31-llms-vs-programming-languages.html

find it amusing that CNN, in an article about capers (of all things), uses scissors wraps and the "Apple mouse" [sic] as examples of bad design.

I mean sure, its a weird design, but there's worse out there!