BLEU Score
Domain: NLP Fundamentals
🎯 Motivation
Measures how close machine translation output is to human reference translations. Higher is better.
📋 Concrete Example
If the model output matches the reference exactly, BLEU = 1.0. Partial matches give partial scores.
🔗 Analogy
Like grading a translation essay by comparing it to the teacher reference answer.