mmT5: Modular Multilingual Pre-Training Solves Source Language Hallucinations.
Jonas PfeifferFrancesco PiccinnoMassimo NicosiaXinyi WangMachel ReidSebastian RuderPublished in: CoRR (2023)
Keyphrases
- cross lingual
- source language
- parallel corpus
- machine translation
- target language
- cross language information retrieval
- cross language
- language independent
- query translation
- training set
- language modeling
- digital libraries
- statistical machine translation
- machine translation system
- supervised learning
- natural language processing
- language model
- training data
- knowledge base