Multilingual Pretraining and Instruction Tuning Improve Cross-Lingual Knowledge Alignment, But Only Shallowly.
Changjiang GaoHongda HuPeng HuJiajun ChenJixing LiShujian HuangPublished in: CoRR (2024)
Keyphrases
- cross lingual
- word alignment
- cross lingual information retrieval
- machine translation
- language independent
- cross language
- language modeling
- multi lingual
- text classification
- knowledge base
- language specific
- parallel corpus
- monolingual and cross lingual
- web news
- translation model
- document clustering
- text categorization
- language model
- knowledge discovery
- cross language information retrieval
- expert systems
- mono lingual
- prior knowledge
- indian languages
- linguistic resources
- text mining
- parallel corpora
- query translation
- event extraction
- natural language processing
- comparable corpora
- knowledge representation
- statistical machine translation
- transfer learning