MINDSET: Energy-based Schema Evolution for Long Conversational Agent Memory
We introduce MINDSET, a memory controller that stores a conversation as immutable episodes and organizes them into versioned schemas through minimum-energy state transitions.
Key points
- Long conversational agents have become essential in our daily lives.
- A useful memory system should preserve both current and historical states, distinguish stale information from active knowledge, retrieve evidence appropriate to the query and avoid repeatedly invoking a large language model to rewrite prior interactions.
- We evaluate MINDSET against 5 memory systems on a reproducible sample of 850 questions (700 LoCoMo + 150 MemoryAgentBench).
- These results show that long-term memory can be better handled as constrained state management rather than continual summarization.
Sources (1)
- [1]MINDSET: Energy-based Schema Evolution for Long Conversational Agent MemoryarXiv (AI, ML, NLP, CV, robotics, multi-agent) · Oct 6, 03:55 PM
We introduce MINDSET, a memory controller that stores a conversation as immutable episodes and organizes them into versioned schemas through minimum-energy state transitions.
Long conversational agents have become essential in our daily lives.
Extractive summary: sentences quoted from the sources.