REMem: Reasoning with Episodic Memory in Language Agents
Yiheng Shu, Saisri Padmaja Jonnalagedda, Xiang Gao, Bernal Jiménez Gutiérrez, Weijian Qi, Kamalika Das, Huan Sun, Yu Su, [2pt] The Ohio State University Intuit AI Research {shu.251, su.809}@osu.edu, {saisri_jonnalagedda, xiang_gao}@intuit.com · 2026
Read REMem: Reasoning with Episodic Memory in Language Agents by Yiheng Shu, Saisri Padmaja Jonnalagedda, Xiang Gao, Bernal Jiménez Gutiérrez, Weijian Qi, Kamalika Das, Huan Sun, Yu Su, [2pt] The Ohio State University Intuit AI Research {shu.251, su.809}@osu.edu, {saisri_jonnalagedda, xiang_gao}@intuit.com (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2602-13530.txt.
Description
People remember concrete episodes in time and place and reason across them. Language agents, by contrast, mostly store semantic facts and struggle to recollect and reason over their own interaction histories. REMem closes that gap with a two-phase design. Offline indexing turns experiences into a hybrid memory graph linking time-aware summaries and facts, and online inference uses an agentic retriever with curated tools to search that graph step by step.
Across four episodic memory benchmarks, REMem beat leading memory systems such as Mem0 and HippoRAG 2. It gained 3.4 percent on recollection and 13.4 percent on reasoning tasks in absolute terms. It also refused unanswerable questions more robustly, a sign of genuine memory rather than fluent guessing. Yiheng Shu leads the work with seven co-authors.
Catalog record
| CREATOR | Yiheng Shu, Saisri Padmaja Jonnalagedda, Xiang Gao, Bernal Jiménez Gutiérrez, Weijian Qi, Kamalika Das, Huan Sun, Yu Su, [2pt] The Ohio State University Intuit AI Research {shu.251, su.809}@osu.edu, {saisri_jonnalagedda, xiang_gao}@intuit.com |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 68 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2602-13530