التفاصيل البيبلوغرافية
| العنوان: |
Towards Modular LLMs by Building and Reusing a Library of LoRAs |
| المؤلفون: |
Ostapenko, Oleksiy, Su, Zhan, Ponti, Edoardo Maria, Charlin, Laurent, Roux, Nicolas Le, Pereira, Matheus, Caccia, Lucas, Sordoni, Alessandro |
| سنة النشر: |
2024 |
| المجموعة: |
Computer Science |
| مصطلحات موضوعية: |
Computer Science - Machine Learning, Computer Science - Computation and Language |
| الوصف: |
The growing number of parameter-efficient adaptations of a base large language model (LLM) calls for studying whether we can reuse such trained adapters to improve performance for new tasks. We study how to best build a library of adapters given multi-task data and devise techniques for both zero-shot and supervised task generalization through routing in such library. We benchmark existing approaches to build this library and introduce model-based clustering, MBC, a method that groups tasks based on the similarity of their adapter parameters, indirectly optimizing for transfer across the multi-task dataset. To re-use the library, we present a novel zero-shot routing mechanism, Arrow, which enables dynamic selection of the most relevant adapters for new inputs without the need for retraining. We experiment with several LLMs, such as Phi-2 and Mistral, on a wide array of held-out tasks, verifying that MBC-based adapters and Arrow routing lead to superior generalization to new tasks. We make steps towards creating modular, adaptable LLMs that can match or outperform traditional joint training. |
| نوع الوثيقة: |
Working Paper |
| URL الوصول: |
http://arxiv.org/abs/2405.11157 |
| رقم الانضمام: |
edsarx.2405.11157 |
| قاعدة البيانات: |
arXiv |