Human-in-the-loop
A design pattern in which a person reviews, approves, or overrides an automated output before it has effect.
Read moreA design pattern in which a person reviews, approves, or overrides an automated output before it has effect.
Read moreUsing a large language model to score or compare outputs, often as a substitute for human evaluation.
Read moreA range around an estimate that expresses uncertainty.
Read moreHow closely predicted confidence matches observed accuracy.
Read moreA structured evaluation of whether a system performs differently across groups, languages, or content types.
Read moreA change in model behaviour or data distribution that degrades performance over time.
Read moreA short document describing a model’s intended use, training data, evaluation results, and limitations.
Read moreDocumentation of a dataset’s provenance, composition, labelling process, and known gaps.
Read moreA confident but unsupported or fabricated output from a language model.
Read moreAdversarial testing designed to find failure modes, misuse paths, and harmful outputs.
Read moreNo terms match that search.