lasu··on Show HN: Parsimony – an open-source benchmark for how much code AI models addThe benchmarks are what allows the AI researchers to improve the models. The AI doesn't get better on its own.