DACL-RAG: Data Augmentation Strategy with Curriculum Learning for Retrieval-Augmented Generation

TL;DR

DACL-RAG introduces a multi-stage training framework combining data augmentation and curriculum learning to improve retrieval-augmented generation, addressing data quality and discriminability issues for better performance.

Contribution

The paper proposes a novel multi-stage training framework with data augmentation and curriculum learning for RAG systems, enhancing training stability and effectiveness.

Findings

Achieves 2-4% performance improvements on four QA datasets.

Effectively addresses data quality and discriminability challenges in RAG training.

Demonstrates consistent gains over existing advanced methods.

Abstract

Retrieval-Augmented Generation (RAG) is an effective method to enhance the capabilities of large language models (LLMs). Existing methods typically optimize the retriever or the generator in a RAG system by directly using the top-k retrieved documents. However, two key issues inherent in the training data constrain the effectiveness of this training paradigm: (1) across different queries, the top-k retrieved documents vary greatly in content quality, with some providing valuable knowledge while others lack critical information or are even misleading, and training on such data in a purely random manner may impair the generator's ability to extract key information; (2) for a given query, the limited set of k documents often exhibits low discriminability, and training solely on them makes it difficult for the retriever to learn how to distinguish between relevant and irrelevant documents. To address these issues, we introduce DACL-RAG, a multi-stage RAG training framework that combines a multi-level Data Augmentation strategy with a multi-stage Curriculum Learning paradigm. The data augmentation strategy constructs comprehensive and diverse training sets with controllable difficulty levels through sample evolution, while the curriculum learning paradigm organizes them into progressive stages for training, ensuring stable and consistent improvements, thereby optimizing the overall performance and generalization of the RAG system more effectively. Our DACL-RAG framework demonstrates consistent effectiveness across four open-domain QA datasets, achieving performance gains of 2% to 4% over multiple advanced methods.