A Balanced Data Approach for Evaluating Cross-Lingual Transfer: Mapping the Linguistic Blood Bank
This work addresses the challenge for developers of multilingual models in optimizing pretraining configurations, though it is incremental as it builds on existing cross-lingual transfer research.
The study tackled the problem of how pretraining language selection affects cross-lingual transfer in BERT-based models by analyzing zero-shot performance under balanced data conditions, identifying donor and recipient languages, and found that their quadratic-time method effectively estimates these relationships across diverse languages and tasks.
We show that the choice of pretraining languages affects downstream cross-lingual transfer for BERT-based models. We inspect zero-shot performance in balanced data conditions to mitigate data size confounds, classifying pretraining languages that improve downstream performance as donors, and languages that are improved in zero-shot performance as recipients. We develop a method of quadratic time complexity in the number of languages to estimate these relations, instead of an exponential exhaustive computation of all possible combinations. We find that our method is effective on a diverse set of languages spanning different linguistic features and two downstream tasks. Our findings can inform developers of large-scale multilingual language models in choosing better pretraining configurations.