The authors implemented Hippocam in an LLM agent and tested it on a suite of eight tasks requiring sustained memory, retrieval of learned knowledge, and adaptation to new situations.
Hippocam enables LLM agents to learn from experience without parameter updates by structuring memory as nested intents and consolidating unused history.
The architecture mimics human memory dynamics by keeping active context focused and recursively abstracting distant experiences while preserving original details for selective recall.
Research Lab
Xiangyi Zeng · Baihang Liu · Xutong Wang · Ze Jin · Yunpeng Li · Qixu Liu
Institute of Information Engineering, Chinese Academy of Sciences · University of Chinese Academy of Sciences
Research Digest··3 min read
The authors propose Hippocam, a memory architecture that organizes an LLM agent's interaction history as a tree of nested intents, consolidating completed tasks and recursively abstracting older unused experiences.
Why this paper
From Institute of Information Engineering, Chinese Academy of Sciences and University of Chinese Academy of Sciences
In one line
An LLM agent can consolidate completed intents into abstract summaries, preserve originals in a tree, and recall details on demand, enabling continual learning without parameter updates.
What we could check
- ·No code link found
- ·No weights link found
- ·No dataset link found
- ·No compute details found
- ·No stated limitations found
- ·No benchmark numbers found
Observed from the paper text and links we have. Absence here means we did not find it, not that it does not exist.
§