Saved in:
Bibliographic Details
Main Authors: Ko, Dayoon, Kim, Jinyoung, Choi, Hahyeon, Kim, Gunhee
Format: Preprint
Published: 2024
Subjects:
Online Access:https://arxiv.org/abs/2406.05606
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866911910702612480
author Ko, Dayoon
Kim, Jinyoung
Choi, Hahyeon
Kim, Gunhee
author_facet Ko, Dayoon
Kim, Jinyoung
Choi, Hahyeon
Kim, Gunhee
contents In the real world, knowledge is constantly evolving, which can render existing knowledge-based datasets outdated. This unreliability highlights the critical need for continuous updates to ensure both accuracy and relevance in knowledge-intensive tasks. To address this, we propose GrowOVER-QA and GrowOVER-Dialogue, dynamic open-domain QA and dialogue benchmarks that undergo a continuous cycle of updates, keeping pace with the rapid evolution of knowledge. Our research indicates that retrieval-augmented language models (RaLMs) struggle with knowledge that has not been trained on or recently updated. Consequently, we introduce a novel retrieval-interactive language model framework, where the language model evaluates and reflects on its answers for further re-retrieval. Our exhaustive experiments demonstrate that our training-free framework significantly improves upon existing methods, performing comparably to or even surpassing continuously trained language models.
format Preprint
id arxiv_https___arxiv_org_abs_2406_05606
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle GrowOVER: How Can LLMs Adapt to Growing Real-World Knowledge?
Ko, Dayoon
Kim, Jinyoung
Choi, Hahyeon
Kim, Gunhee
Computation and Language
In the real world, knowledge is constantly evolving, which can render existing knowledge-based datasets outdated. This unreliability highlights the critical need for continuous updates to ensure both accuracy and relevance in knowledge-intensive tasks. To address this, we propose GrowOVER-QA and GrowOVER-Dialogue, dynamic open-domain QA and dialogue benchmarks that undergo a continuous cycle of updates, keeping pace with the rapid evolution of knowledge. Our research indicates that retrieval-augmented language models (RaLMs) struggle with knowledge that has not been trained on or recently updated. Consequently, we introduce a novel retrieval-interactive language model framework, where the language model evaluates and reflects on its answers for further re-retrieval. Our exhaustive experiments demonstrate that our training-free framework significantly improves upon existing methods, performing comparably to or even surpassing continuously trained language models.
title GrowOVER: How Can LLMs Adapt to Growing Real-World Knowledge?
topic Computation and Language
url https://arxiv.org/abs/2406.05606