Decoding by Contrasting Knowledge: Enhancing Large Language Model Confidence on Edited Facts

ACL ARR 2024 December Submission1749 Authors

Abstract:

The knowledge within large language models (LLMs) may become outdated quickly.
While in-context editing (ICE) is currently the most effective method for knowledge editing (KE), it is constrained by the black-box modeling of LLMs and thus lacks interpretability.
Our work aims to elucidate the superior performance of ICE in KE by analyzing the impacts of in-context new knowledge on token-wise distributions.
We observe that despite a significant boost in logits of the new knowledge, the performance of ICE is still hindered by stubborn knowledge.
We propose a novel approach termed **De**coding by **C**ontrasting **K**nowledge (**DeCK**).
DeCK derives the distribution of the next token by contrasting the logits obtained from the newly edited knowledge guided by ICE with those from the unedited parametric knowledge.
Our experiments demonstrate that DeCK enhances the confidence of LLMs in edited facts.
For instance, it improves the performance of LLaMA3-8B-instruct on MQuAKE by up to 219\%, demonstrating its capability to strengthen ICE.
DeCK can be easily integrated into any ICE method as a decoding component to enhance editing capabilities.

Paper Type: Long

Research Area: Efficient/Low-Resource Methods for NLP

Research Area Keywords: knowledge augmented, knowledge tracing/discovering/inducing, pruning

Contribution Types: Model analysis & interpretability, NLP engineering experiment, Approaches to low-resource settings

Languages Studied: English

Submission Number: 1749