Keywords: nearest neighbors, machine translation
Abstract: We introduce $k$-nearest-neighbor machine translation ($k$NN-MT), which predicts tokens with a nearest-neighbor classifier over a large datastore of cached examples, using representations from a neural translation model for similarity search. This approach requires no additional training and scales to give the decoder direct access to billions of examples at test time, resulting in a highly expressive model that consistently improves performance across many settings. Simply adding nearest-neighbor search improves a state-of-the-art German-English translation model by 1.5 BLEU. $k$NN-MT allows a single model to be adapted to diverse domains by using a domain-specific datastore, improving results by an average of 9.2 BLEU over zero-shot transfer, and achieving new state-of-the-art results---without training on these domains. A massively multilingual model can also be specialized for particular language pairs, with improvements of 3 BLEU for translating from English into German and Chinese. Qualitatively, $k$NN-MT is easily interpretable; it combines source and target context to retrieve highly relevant examples.
One-sentence Summary: We augment the decoder of a pre-trained machine translation model with a nearest neighbor classifier, substantially improving performance in the single language-pair, multilingual and domain adaptation settings, without any additional training.
Code Of Ethics: I acknowledge that I and all co-authors of this work have read and commit to adhering to the ICLR Code of Ethics
Code: [![github](/images/github_icon.svg) urvashik/knnlm](https://github.com/urvashik/knnlm) + [![Papers with Code](/images/pwc_icon.svg) 2 community implementations](https://paperswithcode.com/paper/?openreview=7wCBOfJ8hJM)