Fine-Tuning Transformers: Vocabulary Transfer
This work addresses an incremental improvement for natural language processing practitioners by optimizing fine-tuning efficiency and effectiveness.
The paper tackled the problem of improving fine-tuning performance for transformers by introducing vocabulary transfer, which uses corpus-specific tokenization and initialization strategies, resulting in faster transfer and boosted model performance.
Transformers are responsible for the vast majority of recent advances in natural language processing. The majority of practical natural language processing applications of these models are typically enabled through transfer learning. This paper studies if corpus-specific tokenization used for fine-tuning improves the resulting performance of the model. Through a series of experiments, we demonstrate that such tokenization combined with the initialization and fine-tuning strategy for the vocabulary tokens speeds up the transfer and boosts the performance of the fine-tuned model. We call this aspect of transfer facilitation vocabulary transfer.