Microsoft Research Asia’s Systems for WMT19

By Yingce Xia, Xu Tan, Fei Tian, Fei Gao, Di He, Weicong Chen, Yang Fan, Linyuan Gong, Yichong Leng, Renqian Luo and 5 more. 2019

26 citations recorded by the source corpus.

We Microsoft Research Asia made submissions to 11 language directions in the WMT19 news translation tasks. We won the first place for 8 of the 11 directions and the second place for the other three. Our basic systems are built on Transformer, back translation and knowledge distillation. We integrate several of our rececent techniques to enhance the baseline systems: multi-agent dual learning (MADL), masked sequence-to-sequence pre-training (MASS), neural architecture optimization (NAO), and soft contextual data augmentation (SCA).

Subfields: Artificial Intelligence, Natural Language Processing

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Natural Language Processing

Citations in the corpus

Findings of the 2019 Conference on Machine Translation (WMT19) (2019, methodology): “MSRA.”
Better Document-Level Machine Translation with Bayes’ Rule (2019, methodology): “…of masked sequence-to-sequence pretraining (Song et al., 2019), sequence-level knowledge distillation (Kim and Rush, 2016), and backtranslation (Edunov et al., 2018), the best model from Xia et al. (2019) achieves 30.8, 30.9, and 39.3 on newstest2017, newstest2018, and newstest2019, respectively.”
Putting Machine Translation in Context with the Noisy Channel Model (2019, methodology): “…of masked sequence-to-sequence pretraining (Song et al., 2019), sequence-level knowledge distillation (Kim & Rush, 2016), and backtranslation (Edunov et al., 2018), the best model from (Xia et al., 2019) achieves 30.8, 30.9, and 39.3 on newstest2017, newstest2018, and newstest2019, respectively.”
Selecting, Planning, and Rewriting: A Modular Approach for Data-to-Document Generation and Translation (2019, methodology): “Since we do not have an in-domain mono-lingual German corpus, we ensemble 3 replicas of the ﬁne-tuned MADL models trained with 10, 20, and 30% dropout for the English-German task.”
Using natural language prompts for machine translation (2022, unknown): “In particular, we report the results of Casas et al. (2019) for English-Kazakh, the results of Xia et al. (2019) for German-French and Marie et al. (2019) for German-Czech.”
Findings of the Third Workshop on Neural Generation and Translation (2019, methodology): “As for the MT model, a pre-trained system from (Xia et al., 2019) was fine-tuned on the RotoWire English-German dataset, as well as backtranslated sentences from the original RotoWire dataset for the English-to-German track.”
A Comprehensive Overview of Large Language Models (2023, unknown): “AlexaTM Wikipedia, mC4 [12] NLG: MLSum [331], XSum [332], E2E [333], WebNLG [334]; Machine Translation: Flores-101 [335], English-German WMT’16, English-French WMT’14, German-French WMT’19 [336]; NLP: XNLI [229], XCOPA [252], PAWS-X [231], XWinograd [337], SuperGLUE [3], SQUADv2 [237], MultiArith [3”