All terms
Evaluation
BLEU
A text-overlap metric, common in machine translation, scoring n-gram match to references.
Definition
BLEU is an automatic metric for evaluating generated text, originally designed for machine translation. It scores the overlap of word sequences (n-grams) between a model's output and one or more reference texts, combining these with a penalty for outputs that are too short. Despite weak sensitivity to meaning, it remains widely reported for its simplicity and long history.