Saved in:
Bibliographic Details
Main Authors: Ostapenko, Oleksiy, Su, Zhan, Ponti, Edoardo Maria, Charlin, Laurent, Roux, Nicolas Le, Pereira, Matheus, Caccia, Lucas, Sordoni, Alessandro
Format: Preprint
Published: 2024
Subjects:
Online Access:https://arxiv.org/abs/2405.11157
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866929348267737088
author Ostapenko, Oleksiy
Su, Zhan
Ponti, Edoardo Maria
Charlin, Laurent
Roux, Nicolas Le
Pereira, Matheus
Caccia, Lucas
Sordoni, Alessandro
author_facet Ostapenko, Oleksiy
Su, Zhan
Ponti, Edoardo Maria
Charlin, Laurent
Roux, Nicolas Le
Pereira, Matheus
Caccia, Lucas
Sordoni, Alessandro
contents 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.
format Preprint
id arxiv_https___arxiv_org_abs_2405_11157
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Towards Modular LLMs by Building and Reusing a Library of LoRAs
Ostapenko, Oleksiy
Su, Zhan
Ponti, Edoardo Maria
Charlin, Laurent
Roux, Nicolas Le
Pereira, Matheus
Caccia, Lucas
Sordoni, Alessandro
Machine Learning
Computation and Language
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.
title Towards Modular LLMs by Building and Reusing a Library of LoRAs
topic Machine Learning
Computation and Language
url https://arxiv.org/abs/2405.11157