Sentence Embeddings in NLI with Iterative Refinement Encoders

Tutkimustuotos: ArtikkelijulkaisuArtikkeliTieteellinenvertaisarvioitu

Kuvaus

Sentence-level representations are necessary for various NLP tasks. Recurrent neural networks have proven to be very effective in learning distributed representations and can be trained efficiently on natural language inference tasks. We build on top of one such model and propose a hierarchy of BiLSTM and max pooling layers that implements an iterative refinement strategy and yields state of the art results on the SciTail dataset as well as strong results for SNLI and MultiNLI. We can show that the sentence embeddings learned in this way can be utilized in a wide variety of transfer learning tasks, outperforming InferSent on 7 out of 10 and SkipThought on 8 out of 9 SentEval sentence embedding evaluation tasks. Furthermore, our model beats the InferSent model in 8 out of 10 recently published SentEval probing tasks designed to evaluate sentence embeddings' ability to capture some of the important linguistic properties of sentences.
Alkuperäiskielienglanti
LehtiNatural Language Engineering
Vuosikerta25
Numero4
Sivut467-482
Sivumäärä16
TilaJulkaistu - 31 heinäkuuta 2019
OKM-julkaisutyyppiA1 Alkuperäisartikkeli tieteellisessä aikakauslehdessä, vertaisarvioitu

Tieteenalat

  • 113 Tietojenkäsittely- ja informaatiotieteet
  • 6121 Kielitieteet

Lainaa tätä

@article{f56f5d09183f4a7d868d1fcd4a53e4c0,
title = "Sentence Embeddings in NLI with Iterative Refinement Encoders",
abstract = "Sentence-level representations are necessary for various NLP tasks. Recurrent neural networks have proven to be very effective in learning distributed representations and can be trained efficiently on natural language inference tasks. We build on top of one such model and propose a hierarchy of BiLSTM and max pooling layers that implements an iterative refinement strategy and yields state of the art results on the SciTail dataset as well as strong results for SNLI and MultiNLI. We can show that the sentence embeddings learned in this way can be utilized in a wide variety of transfer learning tasks, outperforming InferSent on 7 out of 10 and SkipThought on 8 out of 9 SentEval sentence embedding evaluation tasks. Furthermore, our model beats the InferSent model in 8 out of 10 recently published SentEval probing tasks designed to evaluate sentence embeddings' ability to capture some of the important linguistic properties of sentences.",
keywords = "113 Computer and information sciences, 6121 Languages",
author = "Talman, {Aarne Johannes} and Anssi Yli-Jyr{\"a} and J{\"o}rg Tiedemann",
year = "2019",
month = "7",
day = "31",
language = "English",
volume = "25",
pages = "467--482",
journal = "Natural Language Engineering",
issn = "1351-3249",
publisher = "Cambridge University Press",
number = "4",

}

Sentence Embeddings in NLI with Iterative Refinement Encoders. / Talman, Aarne Johannes; Yli-Jyrä, Anssi; Tiedemann, Jörg.

julkaisussa: Natural Language Engineering, Vuosikerta 25, Nro 4, 31.07.2019, s. 467-482.

Tutkimustuotos: ArtikkelijulkaisuArtikkeliTieteellinenvertaisarvioitu

TY - JOUR

T1 - Sentence Embeddings in NLI with Iterative Refinement Encoders

AU - Talman, Aarne Johannes

AU - Yli-Jyrä, Anssi

AU - Tiedemann, Jörg

PY - 2019/7/31

Y1 - 2019/7/31

N2 - Sentence-level representations are necessary for various NLP tasks. Recurrent neural networks have proven to be very effective in learning distributed representations and can be trained efficiently on natural language inference tasks. We build on top of one such model and propose a hierarchy of BiLSTM and max pooling layers that implements an iterative refinement strategy and yields state of the art results on the SciTail dataset as well as strong results for SNLI and MultiNLI. We can show that the sentence embeddings learned in this way can be utilized in a wide variety of transfer learning tasks, outperforming InferSent on 7 out of 10 and SkipThought on 8 out of 9 SentEval sentence embedding evaluation tasks. Furthermore, our model beats the InferSent model in 8 out of 10 recently published SentEval probing tasks designed to evaluate sentence embeddings' ability to capture some of the important linguistic properties of sentences.

AB - Sentence-level representations are necessary for various NLP tasks. Recurrent neural networks have proven to be very effective in learning distributed representations and can be trained efficiently on natural language inference tasks. We build on top of one such model and propose a hierarchy of BiLSTM and max pooling layers that implements an iterative refinement strategy and yields state of the art results on the SciTail dataset as well as strong results for SNLI and MultiNLI. We can show that the sentence embeddings learned in this way can be utilized in a wide variety of transfer learning tasks, outperforming InferSent on 7 out of 10 and SkipThought on 8 out of 9 SentEval sentence embedding evaluation tasks. Furthermore, our model beats the InferSent model in 8 out of 10 recently published SentEval probing tasks designed to evaluate sentence embeddings' ability to capture some of the important linguistic properties of sentences.

KW - 113 Computer and information sciences

KW - 6121 Languages

M3 - Article

VL - 25

SP - 467

EP - 482

JO - Natural Language Engineering

JF - Natural Language Engineering

SN - 1351-3249

IS - 4

ER -