Align-to-Distill: Trainable Attention Alignment for Knowledge Distillation in Neural Machine Translation.
Heegon JinSeonil SonJemin ParkYoungseok KimHyungjong NohYeonsoo LeePublished in: LREC/COLING (2024)
Keyphrases
- machine translation
- word alignment
- language independent
- cross lingual
- word level
- information extraction
- target language
- natural language generation
- natural language
- natural language processing
- cross language information retrieval
- language processing
- language resources
- word sense disambiguation
- knowledge representation
- grammar induction
- knowledge base
- statistical machine translation
- machine translation system
- markov networks
- parallel corpora
- knowledge sources
- document retrieval
- information retrieval
- brazilian portuguese
- statistical translation models