| title | MEME: Multi-entity & Evolving Memory Evaluation |
|---|---|
| authors | Seokwon Jung, Alexander Rubinstein, Arnas Uselis, Sangdoo Yun, Seong Joon Oh |
| date | 2026-05-12 |
| venue | arXiv (preprint, not peer reviewed) |
| id | arXiv:2605.12477 |
| affiliation | KAIST AI |
| upvotes | 6 (Hugging Face Daily Papers, 2026-11-01) |
| links | abs pdf hf code |
| primer | Memory for AI agents |
| abstract | LLM-based agents increasingly operate in persistent environments where they must store, update, and reason over information across many sessions. While prior benchmarks evaluate only single-entity updates, MEME defines six tasks spanning the full space defined by the multi-entity and evolving axes, including three not scored by prior work: Cascade and Absence (dependency reasoning) and Deletion (post-removal state). Evaluating six memory systems spanning three memory paradigms on 100 controlled episodes, we find that all systems collapse on dependency reasoning under the default configuration (Cascade: 3%, Absence: 1% in average accuracy) despite adequate static retrieval performance. Prompt optimization, deeper retrieval, reduced filler noise, and most stronger LLMs fail to close this gap. Only a file-based agent paired with Claude Opus 4.7 as its internal LLM partially closes the gap, but at ~70x the baseline cost, indicating closure currently depends on configurations that are not practical at scale. Code and data are available on the project page: https://seokwonjung-jay.github.io/meme-eval/. |
cite
@misc{jung2026meme,
title={MEME: Multi-entity & Evolving Memory Evaluation},
author={Seokwon Jung and Alexander Rubinstein and Arnas Uselis and Sangdoo Yun and Seong Joon Oh},
year={2026},
eprint={2605.12477},
archivePrefix={arXiv},
url={https://arxiv.org/abs/2605.12477}
}