It aggregates reported or independently run scores under a published methodology. Rankings can hide tradeoffs across tasks, cost, latency, safety, and statistical uncertainty, so underlying results matter more than a single position.
A model leaderboard ranks systems according to results on specified evaluations.
It aggregates reported or independently run scores under a published methodology. Rankings can hide tradeoffs across tasks, cost, latency, safety, and statistical uncertainty, so underlying results matter more than a single position.