Skip to main content

Vocabulary of natural language processing

Search from vocabulary

  • Help
align
annotate
download

Concept information

... > NLP methods and tools > model > statistical model > foundation model > language model > mBART
... > NLP methods and tools > statistics > statistical model > foundation model > language model > mBART

Preferred term

mBART  

Definition

  • A sequence-to-sequence denoising auto-encoder pre-trained on large-scale monolingual corpora in many languages using the BART objective. (Liu et al., 2020)

Broader concept

Example

  • Specifically we augment the original vocabulary of mBART with the names of AMR relations and frames occurring at least 5 times in the gold training corpus. (Cai, Li, Ho, Bing & Lam, 2021)
  • The noise function of mBART replaces text spans of arbitrary length with a mask token (35% of the words in each instance) and permutes the order of sentences. (Kasner & Dušek, 2020)
  • Tran et al. (2020) show that fine-tuning mBART using pseudo-parallel data leads to very promising results so we use mBART for our own experiments as well. (Belouadi & Eger, 2023)
  • We base our approach on the mBART model which is pre-trained for multilingual denoising. (Kasner & Dušek, 2020)

In other languages

URI

http://data.loterre.fr/ark:/67375/8LP-PDSXQ6Q0-M

Download this concept:

RDF/XML TURTLE JSON-LD Last modified 4/26/24