Adapter: Parameter-Efficient Transfer Learning for NLP

Authors: Houlsby et al. Year: 2019 ArXiv/Link: https://arxiv.org/abs/1902.00751

Summary

Small bottleneck layers inserted into models enable efficient transfer learning with minimal additional parameters.

Key Concepts

• Adapter modules
• Transfer learning
• Bottleneck architecture
• Parameter sharing
• Lightweight adaptation

Impact

Early approach to parameter-efficient fine-tuning

Category

Fine-tuning & Adaptation

---