OFA: A Framework of Initializing Unseen Subword Embeddings for Efficient Large-scale Multilingual Continued Pretraining
About
Instead of pretraining multilingual language models from scratch, a more efficient method is to adapt existing pretrained language models (PLMs) to new languages via vocabulary extension and continued pretraining. However, this method usually randomly initializes the embeddings of new subwords and introduces substantially more embedding parameters to the model, thus weakening the efficiency. To address these issues, we propose a novel framework: $\textbf{O}$ne $\textbf{F}$or $\textbf{A}$ll ($\textbf{OFA}$), which wisely initializes the embeddings of unseen subwords and thus can adapt a PLM to multiple languages efficiently and effectively. OFA takes advantage of external well-aligned multilingual static word vectors and injects the alignment knowledge into the subword embeddings. In addition, OFA applies matrix factorization and replaces the cumbersome embeddings with two lower-dimensional matrices, which largely reduces the number of parameters. We show OFA accelerates the convergence of continued pretraining, which is environmentally friendly as much fewer carbon footprints are generated. Through extensive experiments, we demonstrate OFA can achieve competitive or better performance than default continued pretraining baselines on a wide range of crosslingual downstream tasks. We make our code and models publicly available.
Related benchmarks
| Task | Dataset | Result | Rank | |
|---|---|---|---|---|
| Question Answering | ARC-E | Accuracy38.17 | 523 | |
| Cross-lingual retrieval | WebFAQ | nDCG@1059.6 | 32 | |
| Text Classification | SIB-200 kmb (test) | Weighted F163.2 | 10 | |
| Text Classification | SIB-200 umb (test) | Weighted F1 (SIB-200 umb test)61.8 | 10 | |
| Text Classification | SIB-200 cjk (test) | Weighted F152.8 | 10 | |
| Text Classification | SIB-200 kon (test) | Weighted F176.9 | 10 | |
| Text Classification | SIB-200 lua (test) | Weighted F168.6 | 10 | |
| Question Answering | Knowledge-based Benchmarks German | ARC Score28.74 | 8 | |
| Question Answering | Knowledge-based Benchmarks Arabic | ARC Score26.52 | 8 | |
| Question Answering | Knowledge-based Benchmarks Vietnamese | ARC Score26.58 | 8 |