Day 2: The model I want is the one that's boring everywhere

Automated excerpt

Previously: Day 0, the benchmark · Day 1, most of my bugs looked like model behaviour The benchmark asks one question twice: can the model do the job, and does it know when it can't? Day 1 ran the local ladder: eight models on a laptop, temperature 0. Day 1's lesson was that most of my bugs looked like model behaviour.

Selected automatically from source text; not independently written or fact-checked. Read the original for full context.