Index Ai Academy
Glossary · Phase 1 · Building with LLMs

BLEU

A score for generated text that counts how many short word sequences match a reference. Higher is closer to the reference.

Try it now

The outlined cell is each row's best match. Their average is the precision.

Learn it properly

BLEU, ROUGE and BERTScore · What the three text-similarity scores reward, how each one misleads, and when a grader or a judge is the better tool. About 35 minutes.

Other terms from the same lesson

  • BERTScore: A score that compares meaning, not exact words, by matching the embeddings of the generated and reference words.
  • ROUGE: A score for summaries that checks how much of the reference's words and word pairs appear in the generated text.
Keep learning

More in plain English.

1286 terms in the glossary, each with the lesson that teaches it.

1286 terms

The glossary

Every word you'll meet in AI and machine learning, explained in everyday language.

Phase 1 · lesson 35 min

BLEU, ROUGE and BERTScore

What the three text-similarity scores reward, how each one misleads, and when a grader or a judge is the better tool.

Never trained an AI? Teach a small one to write, right in your browser. It takes about 20 seconds and needs no account. Teach it to write