3 ms·Any problem set that can make news is implicitly an LLM benchmark.by jprete 2y agoAny problem set that can make news is implicitly an LLM benchmark.