Bahdanau, Dzmitry; Cho, Kyunghyun; Bengio, Yoshua (1 September 2014). Neural Machine Translation by Jointly Learning to Align and Translate. arXiv:1409.0473
Pouget-Abadie, Jean; Bahdanau, Dzmitry; van Merrienboer, Bart; Cho, Kyunghyun; Bengio, Yoshua (October 2014). “Overcoming the Curse of Sentence Length for Neural Machine Translation using Automatic Segmentation”. Proceedings of SSST-8, Eighth Workshop on Syntax, Semantics and Structure in Statistical Translation (Stroudsburg, PA, USA: Association for Computational Linguistics): 78–85. arXiv:1409.1257. doi:10.3115/v1/w14-4009.
Pouget-Abadie, Jean; Bahdanau, Dzmitry; van Merrienboer, Bart; Cho, Kyunghyun; Bengio, Yoshua (October 2014). “Overcoming the Curse of Sentence Length for Neural Machine Translation using Automatic Segmentation”. Proceedings of SSST-8, Eighth Workshop on Syntax, Semantics and Structure in Statistical Translation (Stroudsburg, PA, USA: Association for Computational Linguistics): 78–85. arXiv:1409.1257. doi:10.3115/v1/w14-4009.
“DSL.sk - Sprístupnený nový prekladač postavený na umelej inteligencii, tvrdí že je najlepší” [A new translator based on artificial intelligence has been made available, claims to be the best] (スロバキア語). DSL.sk (2017年8月30日). 2020年7月3日時点のオリジナルよりアーカイブ。2020年8月10日閲覧。 “Prekladač rovnako ako prekladače najlepších existujúcich konkurentov využíva techniku umelej inteligencie a strojového učenia neurónové siete. Podľa tvorcov ale používa inú výrazne vylepšenú architektúru neurónových sietí s inak prepojenými neurónmi ako doterajšie prekladače, čo mu umožňuje prekladať jazyky prirodzenejšie. Detaily technológie zatiaľ nezverejní. (The translator, like the translators of the best existing competitors, uses the technique of artificial intelligence and neural network machine learning. According to the creators, however, it uses a different, significantly improved architecture of neural networks with differently connected neurons than previous translators, which allows it to translate languages more naturally. Details of the technology will not be published yet.)”
Larousserie, David; Leloup, Damien (2017年8月29日). “Quel est le meilleur service de traduction en ligne ? [What is the best online translation service?]” (フランス語). Le Monde. オリジナルの2019年5月31日時点におけるアーカイブ。. https://web.archive.org/web/20190531181754/https://www.lemonde.fr/pixels/article/2017/08/29/quel-est-le-meilleur-service-de-traduction-en-ligne_5177956_4408996.html2019年9月1日閲覧. "Sur un bref extrait d'une tribune, au style sobre d'un ministre, DeepL a également obtenu de meilleurs résultats que les autres services, grâce à des tournures de phrase plus « françaises » et en évitant un petit piège : les quatre autres ont traduit « to discuss these issues further » par « discuter de ces questions plus loin », ce qui n'est pas correct. DeepL a lui correctement traduit par « discuter de ces questions plus en profondeur ». (In a brief excerpt from a podium, in the sober style of a minister, DeepL also obtained better results than the other services, thanks to more "French" turns of phrase and by avoiding a small trap: the other four translated "to discuss these issues further" by "discuter de ces questions plus loin", which is not correct. DeepL correctly translated it as "discuter de ces questions plus en profondeur".)"
“Arriva DeepL, il traduttore automatico che sfida Google [Here comes DeepL, the automatic translator that challenges Google]” (イタリア語). la Repubblica. (2017年8月29日). オリジナルの2019年9月1日時点におけるアーカイブ。. https://web.archive.org/web/20190901133210/https://www.repubblica.it/tecnologia/prodotti/2017/08/29/news/arriva_deepl_il_traduttore_automatico_che_sfida_google-174078830/2019年9月1日閲覧. "Il team della DeepL utilizza questo supercomputer per addestrare le reti di traduzione neurale su una vasta collezione di testi multilingue. Durante l'addestramento, le reti esaminano una moltitudine di traduzioni e imparano a tradurre in maniera indipendente, con una grammatica e una formulazione corrette. ... Nel corso dell'ultimo decennio DeepL ha raccolto oltre un miliardo di testi tradotti di alta qualità, collezionando il miglior materiale d'addestramento possibile per una rete di traduzione neurale. (The DeepL team uses this supercomputer to train neural translation networks on a large collection of multilingual texts. During training, the networks examine a multitude of translations and learn to translate independently, with correct grammar and wording. ... Over the past decade, DeepL has collected over a billion high-quality translated texts, collecting the best possible training material for a neural translation network.)"
“DSL.sk - Sprístupnený nový prekladač postavený na umelej inteligencii, tvrdí že je najlepší” [A new translator based on artificial intelligence has been made available, claims to be the best] (スロバキア語). DSL.sk (2017年8月30日). 2020年7月3日時点のオリジナルよりアーカイブ。2020年8月10日閲覧。 “Prekladač rovnako ako prekladače najlepších existujúcich konkurentov využíva techniku umelej inteligencie a strojového učenia neurónové siete. Podľa tvorcov ale používa inú výrazne vylepšenú architektúru neurónových sietí s inak prepojenými neurónmi ako doterajšie prekladače, čo mu umožňuje prekladať jazyky prirodzenejšie. Detaily technológie zatiaľ nezverejní. (The translator, like the translators of the best existing competitors, uses the technique of artificial intelligence and neural network machine learning. According to the creators, however, it uses a different, significantly improved architecture of neural networks with differently connected neurons than previous translators, which allows it to translate languages more naturally. Details of the technology will not be published yet.)”
Larousserie, David; Leloup, Damien (2017年8月29日). “Quel est le meilleur service de traduction en ligne ? [What is the best online translation service?]” (フランス語). Le Monde. オリジナルの2019年5月31日時点におけるアーカイブ。. https://web.archive.org/web/20190531181754/https://www.lemonde.fr/pixels/article/2017/08/29/quel-est-le-meilleur-service-de-traduction-en-ligne_5177956_4408996.html2019年9月1日閲覧. "Sur un bref extrait d'une tribune, au style sobre d'un ministre, DeepL a également obtenu de meilleurs résultats que les autres services, grâce à des tournures de phrase plus « françaises » et en évitant un petit piège : les quatre autres ont traduit « to discuss these issues further » par « discuter de ces questions plus loin », ce qui n'est pas correct. DeepL a lui correctement traduit par « discuter de ces questions plus en profondeur ». (In a brief excerpt from a podium, in the sober style of a minister, DeepL also obtained better results than the other services, thanks to more "French" turns of phrase and by avoiding a small trap: the other four translated "to discuss these issues further" by "discuter de ces questions plus loin", which is not correct. DeepL correctly translated it as "discuter de ces questions plus en profondeur".)"
“Arriva DeepL, il traduttore automatico che sfida Google [Here comes DeepL, the automatic translator that challenges Google]” (イタリア語). la Repubblica. (2017年8月29日). オリジナルの2019年9月1日時点におけるアーカイブ。. https://web.archive.org/web/20190901133210/https://www.repubblica.it/tecnologia/prodotti/2017/08/29/news/arriva_deepl_il_traduttore_automatico_che_sfida_google-174078830/2019年9月1日閲覧. "Il team della DeepL utilizza questo supercomputer per addestrare le reti di traduzione neurale su una vasta collezione di testi multilingue. Durante l'addestramento, le reti esaminano una moltitudine di traduzioni e imparano a tradurre in maniera indipendente, con una grammatica e una formulazione corrette. ... Nel corso dell'ultimo decennio DeepL ha raccolto oltre un miliardo di testi tradotti di alta qualità, collezionando il miglior materiale d'addestramento possibile per una rete di traduzione neurale. (The DeepL team uses this supercomputer to train neural translation networks on a large collection of multilingual texts. During training, the networks examine a multitude of translations and learn to translate independently, with correct grammar and wording. ... Over the past decade, DeepL has collected over a billion high-quality translated texts, collecting the best possible training material for a neural translation network.)"
Polo, Juan Diego (2017年8月29日). “DeepL, un traductor online que supera al de Google, Microsoft y Facebook” [DeepL, an online translator that outperforms Google, Microsoft and Facebook] (スペイン語). WWWhat's new?. 2019年9月1日時点のオリジナルよりアーカイブ。2019年9月1日閲覧。 “Desde DeepL nos presentan un traductor que quiere revolucionar el mundo de la traducción automática neuronal, ofreciéndolo de forma gratuita en DeepL.com. (DeepL presents us with a translator that wants to revolutionise the world of neural machine translation, offering it for free on DeepL.com.)”
Polo, Juan Diego (2017年8月29日). “DeepL, un traductor online que supera al de Google, Microsoft y Facebook” [DeepL, an online translator that outperforms Google, Microsoft and Facebook] (スペイン語). WWWhat's new?. 2019年9月1日時点のオリジナルよりアーカイブ。2019年9月1日閲覧。 “Desde DeepL nos presentan un traductor que quiere revolucionar el mundo de la traducción automática neuronal, ofreciéndolo de forma gratuita en DeepL.com. (DeepL presents us with a translator that wants to revolutionise the world of neural machine translation, offering it for free on DeepL.com.)”
