Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

AssistRAG: Boosting the Potential of Large Language Models with an Intelligent Information Assistant

Yujia Zhou, Zheng Liu, Zhicheng Dou

TL;DR

Experiments show Assistant-based Retrieval-Augmented Generation significantly outperforms benchmarks, especially benefiting less advanced LLMs, by providing superior reasoning capabilities and accurate responses.

Abstract

The emergence of Large Language Models (LLMs) has significantly advanced natural language processing, but these models often generate factually incorrect information, known as "hallucination". Initial retrieval-augmented generation (RAG) methods like the "Retrieve-Read" framework was inadequate for complex reasoning tasks. Subsequent prompt-based RAG strategies and Supervised Fine-Tuning (SFT) methods improved performance but required frequent retraining and risked altering foundational LLM capabilities. To cope with these challenges, we propose Assistant-based Retrieval-Augmented Generation (AssistRAG), integrating an intelligent information assistant within LLMs. This assistant manages memory and knowledge through tool usage, action execution, memory building, and plan specification. Using a two-phase training approach, Curriculum Assistant Learning and Reinforced Preference Optimization. AssistRAG enhances information retrieval and decision-making. Experiments show AssistRAG significantly outperforms benchmarks, especially benefiting less advanced LLMs, by providing superior reasoning capabilities and accurate responses.

AssistRAG: Boosting the Potential of Large Language Models with an Intelligent Information Assistant

TL;DR

Experiments show Assistant-based Retrieval-Augmented Generation significantly outperforms benchmarks, especially benefiting less advanced LLMs, by providing superior reasoning capabilities and accurate responses.

Abstract

The emergence of Large Language Models (LLMs) has significantly advanced natural language processing, but these models often generate factually incorrect information, known as "hallucination". Initial retrieval-augmented generation (RAG) methods like the "Retrieve-Read" framework was inadequate for complex reasoning tasks. Subsequent prompt-based RAG strategies and Supervised Fine-Tuning (SFT) methods improved performance but required frequent retraining and risked altering foundational LLM capabilities. To cope with these challenges, we propose Assistant-based Retrieval-Augmented Generation (AssistRAG), integrating an intelligent information assistant within LLMs. This assistant manages memory and knowledge through tool usage, action execution, memory building, and plan specification. Using a two-phase training approach, Curriculum Assistant Learning and Reinforced Preference Optimization. AssistRAG enhances information retrieval and decision-making. Experiments show AssistRAG significantly outperforms benchmarks, especially benefiting less advanced LLMs, by providing superior reasoning capabilities and accurate responses.

Paper Structure

This paper contains 28 sections, 2 equations, 5 figures, 8 tables.

Table of Contents

  1. Introduction
  2. Related Work
  3. Retrieval-Augmented Generation
  4. LLM-based Autonomous Agents
  5. Methodology
  6. Task Definition
  7. AssistRAG Overview
  8. Memory Management
  9. Knowledge Management
  10. AssistRAG Training
  11. Curriculum Assistant Learning
  12. Reinforced Preference Optimization
  13. AssistRAG Inference
  14. Experimental Setup
  15. Datasets and Evaluation Metrics
  16. ...and 13 more sections

Figures (5)

  • Figure 1: Comparisons of Naive, Prompt-based, SFT-based and our Assistant-based RAG frameworks.
  • Figure 2: Overview of AssistRAG. AssistRAG enhances LLMs by providing an intelligent information assistant. Endowed with the ability of tool usage, action execution, memory building and plan specification, it can achieve effective memory and knowledge management.
  • Figure 3: Training framework of AssistRAG. It undergoes a two-stage training pipeline through curriculum assistant learning and reinforced preference optimization.
  • Figure 4: The relationships between inference time, cost, and F1 accuracy for different methods.
  • Figure 5: Performance with different training data sizes.