Large language models (LLMs) encode vast amounts of pre-trained knowledge in their parameters, but updating them as real-world information evolves remains a challenge. Existing methodologies and benchmarks primarily target entity substitutions, failing to capture the full breadth of complex real-world dynamics. In this paper, we introduce Knowledge Update Playground (KUP), an automatic pipeline for simulating realistic knowledge updates reflected in an evidence corpora. KUP's evaluation framework includes direct and indirect probes to both test memorization of updated facts and reasoning over them, for any update learning methods. Next, we present a lightweight method called memory conditioned training (MCT), which conditions tokens in the update corpus on self-generated "memory" tokens during training. Our strategy encourages LLMs to surface and reason over newly memorized knowledge at inference. Our results on two strong LLMs show that (1) KUP benchmark is highly challenging, with the best CPT models achieving in indirect probing setting (reasoning) and (2) MCT training significantly outperforms prior continued pre-training (CPT) baselines, improving direct probing (memorization) results by up to .
View on arXiv@article{li2025_2504.12523,
title={ Memorization vs. Reasoning: Updating LLMs with New Knowledge },
author={ Aochong Oliver Li and Tanya Goyal },
journal={arXiv preprint arXiv:2504.12523},
year={ 2025 }
}