Are All Languages Created Equal in Multilingual BERT?
arXiv:2005.09093
Abstract
Multilingual BERT (mBERT) trained on 104 languages has shown surprisingly good cross-lingual performance on several NLP tasks, even without explicit cross-lingual signals. However, these evaluations have focused on cross-lingual transfer with high-resource languages, covering only a third of the languages covered by mBERT. We explore how mBERT performs on a much wider set of languages, focusing on the quality of representation for low-resource languages, measured by within-language performance. We consider three tasks: Named Entity Recognition (99 languages), Part-of-speech Tagging, and Dependency Parsing (54 languages each). mBERT does better than or comparable to baselines on high resource languages but does much worse for low resource languages. Furthermore, monolingual BERT models for these languages do even worse. Paired with similar languages, the performance gap between monolingual BERT and mBERT can be narrowed. We find that better models for low resource languages require more efficient pretraining techniques or more data.
RepL4NLP Workshop 2020 (Best Long Paper)
References in corpus (7)
- Adam: A Method for Stochastic Optimization
- PyTorch: An Imperative Style, High-Performance Deep Learning Library
- On the Cross-lingual Transferability of Monolingual Representations
- Cross-Lingual Ability of Multilingual BERT: An Empirical Study
- Unicoder: A Universal Language Encoder by Pre-training with Multiple Cross-lingual Tasks
- Emerging Cross-lingual Structure in Pretrained Language Models
- Cross-lingual Pre-training Based Transfer for Zero-shot Neural Machine Translation
Cited by in corpus (7)
- Indic-Transformers: An Analysis of Transformer Language Models for Indian Languages
- Robust Optimization for Multilingual Translation with Imbalanced Data
- Unsupervised Translation of German--Lower Sorbian: Exploring Training and Novel Transfer Methods on a Low-Resource Language
- Building Low-Resource NER Models Using Non-Speaker Annotation
- Deriving Disinformation Insights from Geolocalized Twitter Callouts
- Predicting the Performance of Multilingual NLP Models
- Investigating Transfer Learning in Multilingual Pre-trained Language Models through Chinese Natural Language Inference