BLEU Score

BLEU measures semantic similarity using word embeddings; it fails on short texts BLEU measures n-gram precision between generated and reference text; it does not capture meaning, synonyms, or sentence-level fluency BLEU measures recall of reference n-grams; it penalizes long outputs BLEU measures perplexity of a language model; it cannot compare two texts