MEMO: Memory-Augmented Model Context Optimization for Robust Multi-Turn Multi-Agent LLM Games
Yunfei Xie ∗,1 , Kevin Wang ∗,‡,2 , Bobby Cheng ∗,4 , Jianzhu Yao 3 , Zhizhou Sha 2 , Alexander Duffy 5 , Yihan Xi 2 , Hongyuan Mei 6 , Cheston Tan 4 , Chen Wei †,1 , Pramod Viswanath †,3 , Zhangyang Wang †,2 1 Rice University 2 The University of Texas at Austin 3 Princeton University 4 A*STAR 5 Good Start Labs 6 TTIC · 2026
Read MEMO: Memory-Augmented Model Context Optimization for Robust Multi-Turn Multi-Agent LLM Games by Yunfei Xie ∗,1 , Kevin Wang ∗,‡,2 , Bobby Cheng ∗,4 , Jianzhu Yao 3 , Zhizhou Sha 2 , Alexander Duffy 5 , Yihan Xi 2 , Hongyuan Mei 6 , Cheston Tan 4 , Chen Wei †,1 , Pramod Viswanath †,3 , Zhangyang Wang †,2 1 Rice University 2 The University of Texas at Austin 3 Princeton University 4 A*STAR 5 Good Start Labs 6 TTIC (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-2603-09022.txt.
Description
Yunfei Xie and colleagues stabilize multi-agent LLM game rankings with memory-augmented context optimization. Long-horizon play compounds small early deviations across turns and agents, biasing win rates, while prompt choice shifts policies further. Their MEMO couples a persistent memory bank of self-play insights with tournament prompt evolution, lifting mean win rates sharply across five games while cutting run-to-run variance.
Catalog record
| CREATOR | Yunfei Xie ∗,1 , Kevin Wang ∗,‡,2 , Bobby Cheng ∗,4 , Jianzhu Yao 3 , Zhizhou Sha 2 , Alexander Duffy 5 , Yihan Xi 2 , Hongyuan Mei 6 , Cheston Tan 4 , Chen Wei †,1 , Pramod Viswanath †,3 , Zhangyang Wang †,2 1 Rice University 2 The University of Texas at Austin 3 Princeton University 4 A*STAR 5 Good Start Labs 6 TTIC |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 91099 |
| LICENSE | http://creativecommons.org/licenses/by-sa/4.0/ |
| RIGHTS | CC BY SA |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2603.09022