March 5, 20241 citationsOpen Access

Mixture-of-LoRAs: An Efficient Multitask Tuning for Large Language Models

WFWenfeng FengNanfang Hospital CHChuzhan HaoAlibaba Group (China)YZYuewei ZhangNorth China Electric Power University

Key Points

Enhanced performance in multi-task learning prevents catastrophic forgetting and interference.
Experiments show a superior performance compared to traditional tuning methods for large language models.
The Mixture-of-LoRAs architecture combines domain-specific LoRA modules for efficient multitask training and adaptation to new domains with ease and flexibility. Each model adapts quickly to specific domains, supporting diverse language applications across varying tasks.

Abstract

Instruction Tuning has the potential to stimulate or enhance specific capabilities of large language models (LLMs). However, achieving the right balance of data is crucial to prevent catastrophic forgetting and interference between tasks. To address these limitations and enhance training flexibility, we propose the Mixture-of-LoRAs (MoA) architecture which is a novel and parameter-efficient tuning method designed for multi-task learning with LLMs. In this paper, we start by individually training multiple domain-specific LoRA modules using corresponding supervised corpus data. These LoRA modules can be aligned with the expert design principles observed in Mixture-of-Experts (MoE). Subsequently, we combine the multiple LoRAs using an explicit routing strategy and introduce domain labels to facilitate multi-task learning, which help prevent interference between tasks and ultimately enhances the performance of each individual task. Furthermore, each LoRA model can be iteratively adapted to a new domain, allowing for quick domain-specific adaptation. Experiments on diverse tasks demonstrate superior and robust performance, which can further promote the wide application of domain-specific LLMs.

AI에게 질문

Bookmark

View Full Paper