AVocaDo: Strategy for Adapting Vocabulary to Downstream Domain
This addresses the issue of vocabulary mismatch in fine-tuning for domain-specific NLP tasks, offering an incremental improvement over standard transfer learning methods.
The paper tackles the problem of suboptimal pretrained vocabularies in transfer learning when domain discrepancies exist, by proposing to adapt the vocabulary to downstream domains and achieving consistent performance improvements across diverse domains.
During the fine-tuning phase of transfer learning, the pretrained vocabulary remains unchanged, while model parameters are updated. The vocabulary generated based on the pretrained data is suboptimal for downstream data when domain discrepancy exists. We propose to consider the vocabulary as an optimizable parameter, allowing us to update the vocabulary by expanding it with domain-specific vocabulary based on a tokenization statistic. Furthermore, we preserve the embeddings of the added words from overfitting to downstream data by utilizing knowledge learned from a pretrained language model with a regularization term. Our method achieved consistent performance improvements on diverse domains (i.e., biomedical, computer science, news, and reviews).