Carpe Diem: On the Evaluation of World Knowledge in Lifelong Language Models
This work addresses the challenge of evaluating and improving language models for real-world knowledge updates, which is incremental as it builds on existing continual learning methods.
The authors tackled the problem of language models adapting to evolving world knowledge by introducing EvolvingQA, a temporally evolving question-answering benchmark, and found that existing continual learning baselines struggle with updating and removing outdated knowledge, particularly for numerical or temporal information.
The dynamic nature of knowledge in an ever-changing world presents challenges for language models trained on static data; the model in the real world often requires not only acquiring new knowledge but also overwriting outdated information into updated ones. To study the ability of language models for these time-dependent dynamics in human language, we introduce a novel task, EvolvingQA, a temporally evolving question-answering benchmark designed for training and evaluating LMs on an evolving Wikipedia database. The construction of EvolvingQA is automated with our pipeline using large language models. We uncover that existing continual learning baselines suffer from updating and removing outdated knowledge. Our analysis suggests that models fail to rectify knowledge due to small weight gradients. In addition, we elucidate that language models particularly struggle to reflect the change of numerical or temporal information. Our work aims to model the dynamic nature of real-world information, suggesting faithful evaluations of the evolution-adaptability of language models.