Is this due to Google wanting to catch up on benchmarks at all costs, and trains their models to be overly meticulous to do so?
I'd guess that it's because Google eats its own dogfood, and if they didn't train the models to be overly meticulous, the existing codebase would implode under a tsunami of slop.
I'd guess that it's because Google eats its own dogfood, and if they didn't train the models to be overly meticulous, the existing codebase would implode under a tsunami of slop.