Open Access

BPEmb: Pre-trained Subword Embeddings in 275 Languages (LREC 2018)

Thumbnail Image

Date

Type

Authors

Heinzerling, Benjamin

Journal Title

Journal ISSN

Volume Title

Publisher

heiDATA

Abstract

Description

BPEmb is a collection of pre-trained subword unit embeddings in 275 languages, based on Byte-Pair Encoding (BPE). In an evaluation using fine-grained entity typing as testbed, BPEmb performs competitively, and for some languages better than alternative subword approaches, while requiring vastly fewer resources and no tokenization.

Keywords

Computer and Information Science, subword embeddings, byte-pair encoding, multilingual

Citation

Identifier

Endorsement

DFG Classification

Project(s)

Faculty

Collections

AIPHES Heidelberg

License

license.name.undefined
Except where otherwise noted, this license is described as license.name.undefined