English

Multilingual Universal Sentence Encoder for Semantic Retrieval

Computation and Language 2019-07-10 v1

Abstract

We introduce two pre-trained retrieval focused multilingual sentence encoding models, respectively based on the Transformer and CNN model architectures. The models embed text from 16 languages into a single semantic space using a multi-task trained dual-encoder that learns tied representations using translation based bridge tasks (Chidambaram al., 2018). The models provide performance that is competitive with the state-of-the-art on: semantic retrieval (SR), translation pair bitext retrieval (BR) and retrieval question answering (ReQA). On English transfer learning tasks, our sentence-level embeddings approach, and in some cases exceed, the performance of monolingual, English only, sentence embedding models. Our models are made available for download on TensorFlow Hub.

Keywords

Cite

@article{arxiv.1907.04307,
  title  = {Multilingual Universal Sentence Encoder for Semantic Retrieval},
  author = {Yinfei Yang and Daniel Cer and Amin Ahmad and Mandy Guo and Jax Law and Noah Constant and Gustavo Hernandez Abrego and Steve Yuan and Chris Tar and Yun-Hsuan Sung and Brian Strope and Ray Kurzweil},
  journal= {arXiv preprint arXiv:1907.04307},
  year   = {2019}
}

Comments

6 pages, 6 tables, 2 listings, and 1 figure

R2 v1 2026-06-23T10:16:33.821Z