HiMem: Hierarchical Long-Term Memory for LLM Long-Horizon Agents
Ningning Zhang, Xingxing Yang, Zhizhong Tan, Weiping Deng, Wenyong Wang · 2026
Read HiMem: Hierarchical Long-Term Memory for LLM Long-Horizon Agents by Ningning Zhang, Xingxing Yang, Zhizhong Tan, Weiping Deng, Wenyong Wang (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-2601-06377.txt.
Description
An agent that chats for months needs a mind store that can grow and fix its own flaws. Ningning Zhang and 4 co-authors built HiMem for that need in a 2026 paper. Past stores have a hard time with change, scale, and self-fixes in long talks.
HiMem has two tiers. One tier holds tales of past chats, cut at points where the topic shifts or a surprise lands. The next tier holds firm facts, drawn from those tales in a few clean steps. The two tiers link by sense, so a search can jump from tale to fact and back. When new facts clash with old ones, a fix step heals the store. On long-chat tests, HiMem beat the main past tools on truth, fit, and thought, while its cost stayed low.
Catalog record
| CREATOR | Ningning Zhang, Xingxing Yang, Zhizhong Tan, Weiping Deng, Wenyong Wang |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 58 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2601-06377