Zobrazeno 1 - 10
of 16
pro vyhledávání: '"Minh-Thang Luong"'
Publikováno v:
EMNLP (1)
We introduce Electric, an energy-based cloze model for representation learning over text. Like BERT, it is a conditional generative model of tokens given their contexts. However, Electric does not use masking or output a full distribution over tokens
Externí odkaz:
https://explore.openaire.eu/search/publication?articleId=doi_dedup___::09cbefd5bd3eb772bcb486ffe12442f2
Publikováno v:
ACL (1)
It can be challenging to train multi-task neural networks that outperform or even match their single-task counterparts. To help address this, we propose using knowledge distillation where single-task models teach a multi-task model. We enhance this t
Externí odkaz:
https://explore.openaire.eu/search/publication?articleId=doi_dedup___::1aced47e53fe5c86cf6c6d7501ad32f1
Publikováno v:
NMT@ACL
This document describes the findings of the Second Workshop on Neural Machine Translation and Generation, held in concert with the annual conference of the Association for Computational Linguistics (ACL 2018). First, we summarize the research trends
Externí odkaz:
https://explore.openaire.eu/search/publication?articleId=doi_dedup___::4ae952df1ef66267d8033f90c9daf230
https://doi.org/10.18653/v1/w18-2701
https://doi.org/10.18653/v1/w18-2701
Publikováno v:
EMNLP
Unsupervised representation learning algorithms such as word2vec and ELMo improve the accuracy of many supervised NLP models, mainly because they can take advantage of large amounts of unlabeled text. However, the supervised models only learn from ta
Externí odkaz:
https://explore.openaire.eu/search/publication?articleId=doi_dedup___::0c5f0f3b53aed3ea8ce03d10a8e6bfd4
Publikováno v:
EMNLP
The standard content-based attention mechanism typically used in sequence-to-sequence models is computationally expensive as it requires the comparison of large encoder and decoder states at each time step. In this work, we propose an alternative att
Externí odkaz:
https://explore.openaire.eu/search/publication?articleId=doi_dedup___::c6de98be4fa910348737f8f56e9198fb
http://arxiv.org/abs/1707.00110
http://arxiv.org/abs/1707.00110
Publikováno v:
Proceedings of the 2017 Conference on Empirical Methods in Natural Language Processing.
Neural Machine Translation (NMT) has shown remarkable progress over the past few years with production systems now being deployed to end-users. One major drawback of current architectures is that they are expensive to train, typically requiring days
Externí odkaz:
https://explore.openaire.eu/search/publication?articleId=doi_dedup___::7c6c0faaa101e0c8cebc2e072996eac7
https://doi.org/10.18653/v1/d17-1151
https://doi.org/10.18653/v1/d17-1151
Publikováno v:
Transactions of the Association for Computational Linguistics. 1:315-326
Grounded language learning, the task of mapping from natural language to a representation of meaning, has attracted more and more interest in recent years. In most work on this topic, however, utterances in a conversation are treated independently an
Externí odkaz:
https://explore.openaire.eu/search/publication?articleId=doi_________::4960390ea522406f6d45780670cfacbc
https://doi.org/10.1162/tacl_a_00230
https://doi.org/10.1162/tacl_a_00230
Publikováno v:
ACL (1)
Nearly all previous work on neural machine translation (NMT) has used quite restricted vocabularies, perhaps with a subsequent method to patch in unknown words. This paper presents a novel word-character solution to achieving open vocabulary NMT. We
Externí odkaz:
https://explore.openaire.eu/search/publication?articleId=doi_dedup___::eba582e2004230a7cfd00ef13faa35d7
http://arxiv.org/abs/1604.00788
http://arxiv.org/abs/1604.00788
Publikováno v:
ACL (1)
Externí odkaz:
https://explore.openaire.eu/search/publication?articleId=doi_________::16618599b8d0361b34202c004f64b7e3
https://doi.org/10.18653/v1/p16-1007
https://doi.org/10.18653/v1/p16-1007
Publikováno v:
CoNLL
Neural Machine Translation (NMT), like many other deep learning domains, typically suffers from over-parameterization, resulting in large storage sizes. This paper examines three simple magnitude-based pruning schemes to compress NMT models, namely c
Externí odkaz:
https://explore.openaire.eu/search/publication?articleId=doi_dedup___::56189167356fdc4c068d0ad236ae3139