All termsTry it in the Evaluation Playground labExplore in the AI Universe
Glossary · Term
Evaluation Metrics
L3 · AdvancedFoundational
In plain terms
Numbers that measure how well a system performs.
Technical definition
Quantities computed from predictions and ground truth (e.g. precision, recall, F1) or from human/LLM judgments, used to compare systems and detect regressions; the right metric depends on the task and its error costs.
Example
Reporting precision and recall, not just accuracy, on an imbalanced dataset.