Our new X account is live! Follow @wizwand_team for updates
WorkDL logo mark

Supervised Learning of Universal Sentence Representations from Natural Language Inference Data

About

Many modern NLP systems rely on word embeddings, previously trained in an unsupervised manner on large corpora, as base features. Efforts to obtain embeddings for larger chunks of text, such as sentences, have however not been so successful. Several attempts at learning unsupervised representations of sentences have not reached satisfactory enough performance to be widely adopted. In this paper, we show how universal sentence representations trained using the supervised data of the Stanford Natural Language Inference datasets can consistently outperform unsupervised methods like SkipThought vectors on a wide range of transfer tasks. Much like how computer vision uses ImageNet to obtain features, which can then be transferred to other tasks, our work tends to indicate the suitability of natural language inference for transfer learning to other NLP tasks. Our encoder is publicly available.

Alexis Conneau, Douwe Kiela, Holger Schwenk, Loic Barrault, Antoine Bordes• 2017

Related benchmarks

TaskDatasetResultRank
Natural Language InferenceSNLI (test)
Accuracy85.3
681
Semantic Textual SimilaritySTS tasks (STS12, STS13, STS14, STS15, STS16, STS-B, SICK-R) various (test)
STS12 Score52.86
393
Sentiment ClassificationCR
Accuracy86.3
142
Sentiment ClassificationMR (test)
Accuracy81.1
142
Subjectivity ClassificationSubj (test)
Accuracy93.7
125
Text-to-Image RetrievalMSCOCO (1K test)
R@133.9
104
Natural Language InferenceSciTail (test)
Accuracy85.1
86
Semantic Textual SimilaritySTS-B
Spearman's Rho (x100)68.03
70
Caption RetrievalMS COCO Karpathy 1k (test)
R@142.6
62
Natural Language InferenceMultiNLI Mismatched
Accuracy73
60
Showing 10 of 48 rows

Other info

Code

Follow for update