Nearest Neighbor Knowledge Distillation for Neural Machine Translation

TL;DR

This paper introduces Nearest Neighbor Knowledge Distillation (NN-KD), a method that precomputes nearest neighbor information to improve neural machine translation without increasing inference time.

Contribution

The paper proposes moving NN search to preprocessing and training an NMT model to directly incorporate NN knowledge, reducing inference costs.

Findings

Achieves consistent improvements over NN-MT baselines

Maintains same training and decoding speed as standard NMT

Addresses overcorrection in translation outputs

Abstract

k-nearest-neighbor machine translation (NN-MT), proposed by Khandelwal et al. (2021), has achieved many state-of-the-art results in machine translation tasks. Although effective, NN-MT requires conducting NN searches through the large datastore for each decoding step during inference, prohibitively increasing the decoding cost and thus leading to the difficulty for the deployment in real-world applications. In this paper, we propose to move the time-consuming NN search forward to the preprocessing phase, and then introduce Nearest Neighbor Knowledge Distillation (NN-KD) that trains the base NMT model to directly learn the knowledge of NN. Distilling knowledge retrieved by NN can encourage the NMT model to take more reasonable target tokens into consideration, thus addressing the overcorrection problem. Extensive experimental results show that, the proposed method achieves consistent improvement over the state-of-the-art baselines including NN-MT, while maintaining the same training and decoding speed as the standard NMT model.