Revisiting Source Context in Nearest Neighbor Machine Translation
Nearest neighbor machine translation (k NN-MT), which interpolates target token
probabilities with estimates derived from additional examples, has achieved significant …
probabilities with estimates derived from additional examples, has achieved significant …
Nearest neighbor machine translation is meta-optimizer on output projection layer
Nearest Neighbor Machine Translation ($ k $ NN-MT) has achieved great success in
domain adaptation tasks by integrating pre-trained Neural Machine Translation (NMT) …
domain adaptation tasks by integrating pre-trained Neural Machine Translation (NMT) …
INK: Injecting KNN knowledge in nearest neighbor machine translation
Neural machine translation has achieved promising results on many translation tasks.
However, previous studies have shown that neural models induce a non-smooth …
However, previous studies have shown that neural models induce a non-smooth …
The Unreasonable Effectiveness of Random Target Embeddings for Continuous-Output Neural Machine Translation
E Tokarchuk, V Niculae - arXiv preprint arXiv:2310.20620, 2023 - arxiv.org
Continuous-output neural machine translation (CoNMT) replaces the discrete next-word
prediction problem with an embedding prediction. The semantic structure of the target …
prediction problem with an embedding prediction. The semantic structure of the target …
An Enhanced Method for Mongolian-Chinese Neural Machine Translation Using Multilingual Datastores and Chinese-Centric Methods
B Wang, Y Ji, N Wu, X Liu, Y Wang, R Mao… - … Conference on Natural …, 2024 - Springer
Low-resource language translation remains a significant challenge in natural language
processing, particularly for the Mongolian-Chinese language pair under the “Belt and Road” …
processing, particularly for the Mongolian-Chinese language pair under the “Belt and Road” …