LLM-based Agent Memory Module

Context:
- It can (typically) support Data Reading Operations, Data Writing Operations, and Reflection Operations.
- It can (typically) be referenced in an LLM-based Agent Architecture.
- It can consist of Short-Term Memory Store and a Long-Term Memory Store.
- It can utilize various Data Formats such as Text-based Memory, Vector-based Memory, and Database-based Memory.
- It can contribute to Agent Consistency, Agent Context-Awareness, and Agent Knowledge Retention.
- ...
Example(s):
- In a Customer Service Bot, it could store previous customer queries and solutions, thereby allowing the agent to offer more personalized and efficient responses.
- In an Autonomous Vehicle, it could store data on previously encountered road conditions, improving the vehicle's handling in similar future scenarios.
- The long-term store in GITM [Zhong et al., 2023].
- The hybrid memory in Voyager Agent [Jiang et al., 2022].
- ...
Counter-Example(s):
- a Stateless ML Model, like BERT with no external memory.
See: Memory System, RAG Algorithm.

References

(Wang, Ma et al., 2023) ⇒ Lei Wang, Chen Ma, Xueyang Feng, ..., and Ji-Rong Wen. (2023). “A Survey on Large Language Model based Autonomous Agents.” In: arXiv preprint arXiv:2308.11432. doi:10.48550/arXiv.2308.11432
- QUOTE: The memory module plays a very important role in the construction of AI agents. It stores information perceived from the environment and leverages the recorded memories to facilitate future actions. The memory module can help the agent to accumulate experiences, self-evolve, and behave in a more consistent, reasonable, and effective manner. This section provides a comprehensive overview of the memory module, focusing on its structures, formats, and operations.
- Figure 2: A unified framework for the architecture design of LLM-based autonomous AI agent.