Advanced NLP Interview Questions, issue 25, Dec 30, 2025

The Back-Translation Direction Trap

Senior NLP Engineer interview at Google DeepMind, and the interviewer asks:

β€œWe need to improve our π˜‘π˜’π˜±π˜’π˜―π˜¦π˜΄π˜¦-𝘡𝘰-𝘌𝘯𝘨𝘭π˜ͺ𝘴𝘩 translation model. We have 10k parallel pairs and 1 billion lines of monolingual English text. To use 𝐁𝐚𝐜𝐀-π“π«πšπ§π¬π₯𝐚𝐭𝐒𝐨𝐧 effectively, which direction do we generate data, and exactly how do we pair it for training?”

Don’t say: β€œWe translate the English data into Japanese to check if the model is consistent.”

Why generating synthetic sources (not targets) is the only way to preserve decoder fluency in production NMT systems.

The full answer, with the mechanism and the arithmetic, is for paid subscribers on Substack.

Read it on Substack

Get the next one

Free on Substack. Unsubscribe in one click.