Fine-Tuning Small Embeddings for Elevated Performance
This addresses data scarcity for low-resource language NLP, but is incremental as it applies existing fine-tuning methods to a specific case.
The paper tackles the problem of limited data for contextual embeddings in low-resource languages like Nepali by fine-tuning an incomplete BERT model on unseen data, showing that this approach drastically improves results compared to the original baseline, though an oracle model remains better on average.
Contextual Embeddings have yielded state-of-the-art results in various natural language processing tasks. However, these embeddings are constrained by models requiring large amounts of data and huge computing power. This is an issue for low-resource languages like Nepali as the amount of data available over the internet is not always sufficient for the models. This work has taken an incomplete BERT model with six attention heads pretrained on Nepali language and finetuned it on previously unseen data. The obtained results from intrinsic and extrinsic evaluations have been compared to the results drawn from the original model baseline and a complete BERT model pretrained on Nepali language as the oracle. The results demonstrate that even though the oracle is better on average, finetuning the small embeddings drastically improves results compared to the original baseline.