@inproceedings{Lample2019,
abstract = {Recent studies have demonstrated the efficiency of generative pretraining for English natural language understanding. In this work, we extend this approach to multiple languages and show the effectiveness of cross-lingual pretraining. We propose two methods to learn cross-lingual language models (XLMs): one unsupervised that only relies on monolingual data, and one supervised that leverages parallel data with a new cross-lingual language model objective. We obtain state-of-the-art results on cross-lingual classification, unsupervised and supervised machine translation. On XNLI, our approach pushes the state of the art by an absolute gain of 4.9% accuracy. On unsupervised machine translation, we obtain 34.3 BLEU on WMT'16 German-English, improving the previous state of the art by more than 9 BLEU. On supervised machine translation, we obtain a new state of the art of 38.5 BLEU on WMT'16 Romanian-English, outperforming the previous best approach by more than 4 BLEU. Our code and pretrained models will be made publicly available.},
archivePrefix = {arXiv},
arxivId = {1901.07291},
author = {Lample, Guillaume and Conneau, Alexis},
booktitle = {33rd Conference on Neural Information Processing Systems (NeurIPS 2019)},
eprint = {1901.07291},
file = {:Users/shanest/Documents/Library/Lample, Conneau/33rd Conference on Neural Information Processing Systems (NeurIPS 2019)/Lample, Conneau - 2019 - Cross-lingual Language Model Pretraining.pdf:pdf},
keywords = {method: cross-linguistic,model},
title = {{Cross-lingual Language Model Pretraining}},
url = {http://arxiv.org/abs/1901.07291},
year = {2019}
}

{"_id":"tJv4wtTE6rswsSbGE","bibbaseid":"lample-conneau-crosslinguallanguagemodelpretraining-2019","authorIDs":[],"author_short":["Lample, G.","Conneau, A."],"bibdata":{"bibtype":"inproceedings","type":"inproceedings","abstract":"Recent studies have demonstrated the efficiency of generative pretraining for English natural language understanding. In this work, we extend this approach to multiple languages and show the effectiveness of cross-lingual pretraining. We propose two methods to learn cross-lingual language models (XLMs): one unsupervised that only relies on monolingual data, and one supervised that leverages parallel data with a new cross-lingual language model objective. We obtain state-of-the-art results on cross-lingual classification, unsupervised and supervised machine translation. On XNLI, our approach pushes the state of the art by an absolute gain of 4.9% accuracy. On unsupervised machine translation, we obtain 34.3 BLEU on WMT'16 German-English, improving the previous state of the art by more than 9 BLEU. On supervised machine translation, we obtain a new state of the art of 38.5 BLEU on WMT'16 Romanian-English, outperforming the previous best approach by more than 4 BLEU. Our code and pretrained models will be made publicly available.","archiveprefix":"arXiv","arxivid":"1901.07291","author":[{"propositions":[],"lastnames":["Lample"],"firstnames":["Guillaume"],"suffixes":[]},{"propositions":[],"lastnames":["Conneau"],"firstnames":["Alexis"],"suffixes":[]}],"booktitle":"33rd Conference on Neural Information Processing Systems (NeurIPS 2019)","eprint":"1901.07291","file":":Users/shanest/Documents/Library/Lample, Conneau/33rd Conference on Neural Information Processing Systems (NeurIPS 2019)/Lample, Conneau - 2019 - Cross-lingual Language Model Pretraining.pdf:pdf","keywords":"method: cross-linguistic,model","title":"Cross-lingual Language Model Pretraining","url":"http://arxiv.org/abs/1901.07291","year":"2019","bibtex":"@inproceedings{Lample2019,\nabstract = {Recent studies have demonstrated the efficiency of generative pretraining for English natural language understanding. In this work, we extend this approach to multiple languages and show the effectiveness of cross-lingual pretraining. We propose two methods to learn cross-lingual language models (XLMs): one unsupervised that only relies on monolingual data, and one supervised that leverages parallel data with a new cross-lingual language model objective. We obtain state-of-the-art results on cross-lingual classification, unsupervised and supervised machine translation. On XNLI, our approach pushes the state of the art by an absolute gain of 4.9% accuracy. On unsupervised machine translation, we obtain 34.3 BLEU on WMT'16 German-English, improving the previous state of the art by more than 9 BLEU. On supervised machine translation, we obtain a new state of the art of 38.5 BLEU on WMT'16 Romanian-English, outperforming the previous best approach by more than 4 BLEU. Our code and pretrained models will be made publicly available.},\narchivePrefix = {arXiv},\narxivId = {1901.07291},\nauthor = {Lample, Guillaume and Conneau, Alexis},\nbooktitle = {33rd Conference on Neural Information Processing Systems (NeurIPS 2019)},\neprint = {1901.07291},\nfile = {:Users/shanest/Documents/Library/Lample, Conneau/33rd Conference on Neural Information Processing Systems (NeurIPS 2019)/Lample, Conneau - 2019 - Cross-lingual Language Model Pretraining.pdf:pdf},\nkeywords = {method: cross-linguistic,model},\ntitle = {{Cross-lingual Language Model Pretraining}},\nurl = {http://arxiv.org/abs/1901.07291},\nyear = {2019}\n}\n","author_short":["Lample, G.","Conneau, A."],"key":"Lample2019","id":"Lample2019","bibbaseid":"lample-conneau-crosslinguallanguagemodelpretraining-2019","role":"author","urls":{"Paper":"http://arxiv.org/abs/1901.07291"},"keyword":["method: cross-linguistic","model"],"metadata":{"authorlinks":{}},"downloads":0},"bibtype":"inproceedings","biburl":"https://www.shane.st/teaching/575/win20/MachineLearning-interpretability.bib","creationDate":"2020-01-06T20:16:55.585Z","downloads":0,"keywords":["method: cross-linguistic","model"],"search_terms":["cross","lingual","language","model","pretraining","lample","conneau"],"title":"Cross-lingual Language Model Pretraining","year":2019,"dataSources":["okYcdTpf4JJ2zkj7A"]}