Parenting: Optimizing Knowledge Selection of Retrieval-Augmented Language Models with Parameter Decoupling and Tailored Tuning

TL;DR

Parenting is a new framework for retrieval-augmented language models that decouples and optimizes internal parameters to better control knowledge adherence and robustness, improving performance across datasets.

Contribution

It introduces a parameter decoupling and tailored tuning strategy inspired by human cognition, enhancing knowledge control in retrieval-augmented models.

Findings

Improves knowledge adherence and robustness in RAG models

Demonstrates effectiveness across multiple datasets and models

Provides a generalizable framework for parameter optimization

Abstract

Retrieval-Augmented Generation (RAG) offers an effective solution to the issues faced by Large Language Models (LLMs) in hallucination generation and knowledge obsolescence by incorporating externally retrieved knowledge. However, existing methods lack effective control mechanisms for integrating internal and external knowledge. Inspired by human cognitive processes, we propose Parenting, a novel framework that decouples, identifies, and purposefully optimizes parameter subspaces related to adherence and robustness. Specifically, Parenting utilizes a key parameter mining method that combines forward and backward propagation signals to localize subspaces representing different capabilities. Then, Parenting employs a type-tailored tuning strategy, applying specific and appropriate optimizations to different subspaces, aiming to achieve a balanced enhancement of both adherence and robustness. Extensive experiments on various datasets and models validate the effectiveness and generalizability of our method.