MultiFiT: Efficient Multi-lingual Language Model Fine-tuning
This enables practitioners to train language models efficiently for low-resource languages, though it is incremental as it builds on existing fine-tuning and cross-lingual approaches.
The paper tackles the problem of efficiently training language models for low-resource languages, proposing MultiFiT and a zero-shot method that outperform models trained with much more data and compute on cross-lingual classification datasets.
Pretrained language models are promising particularly for low-resource languages as they only require unlabelled data. However, training existing models requires huge amounts of compute, while pretrained cross-lingual models often underperform on low-resource languages. We propose Multi-lingual language model Fine-Tuning (MultiFiT) to enable practitioners to train and fine-tune language models efficiently in their own language. In addition, we propose a zero-shot method using an existing pretrained cross-lingual model. We evaluate our methods on two widely used cross-lingual classification datasets where they outperform models pretrained on orders of magnitude more data and compute. We release all models and code.