AI Dictionary › Fondamenti AI
Classifica di valutazione (Leaderboard)
A leaderboard is a public ranking that orders AI models by the scores they achieve on one or more benchmarks. It lets researchers, companies and developers quickly compare the performance of different models on common tasks, such as language understanding, mathematical reasoning or code generation. It is a widely used reference point for navigating the dozens of available models.
Each leaderboard defines its own rules: which tests it includes, how scores are weighted, and how often it is updated. Some rely on standardized automated tests, others incorporate preference votes collected from real users comparing responses from different models. Scores are recalculated as new models or new versions of existing ones are released.
Leaderboards are consulted by those choosing a model for a product, by researchers positioning their work against the state of the art, and by teams communicating a new model's progress to the public. They have become a common reference point in debates over which system is most advanced at a given moment.
AI leaderboards spread alongside the proliferation of language models, when it became useful to have shared, updatable reference points instead of isolated comparisons between just two systems.
Grace shows internal rankings to its users: the scores earned in scenarios are compared with those of the community, much like a leaderboard, to encourage continuous improvement.
From our network
AGORÀ Intelligence: Enterprise AI Governance Platform
Govern AI at scale: policies, adoption and measurable results on your data. Built for boards and C-suite.
Visit agora-intelligence.com →From the Agora Intelligence blog
📱 Download the Android app (beta) iOS coming soon
Say what you mean. Get what you need.
Grace Certified, the AI coach that trains and certifies your prompt engineering, by Agora Intelligence.