Scholar Feed

Agent / long-term memory

MemoryBank

MemoryBank: Enhancing Large Language Models with Long-Term Memory

Superseded baseline#14 of 63 most-superseded · first seen May 17, 2023

Superseded — cited as a baseline and beaten by newer methods

3 papers critique it · 1 beat it on benchmarks

What papers say

Verbatim critique sentences, each from a paper that cites MemoryBank as a baseline.

Existing memory systems~memgpt,memorybank,smolagents,agentlite for LLM agents provide basic memory storage functionality. These systems require agent developers to predefine memory storage structures, specify storage points within the workflow, and establish retrieval timing.
— A-MEM: Agentic Memory for LLM Agents
“Early memory mechanisms in LLM-based agents typically relied on heuristic-based static workflows.”
— AtomMem : Learnable Dynamic Agentic Memory with Atomic Memory Operation
“existing benchmarks~MemoryBank,locomo abstract away the memory formation process during interaction and do not align with this view.”
— EMemBench: Interactive Benchmarking of Episodic Memory for VLM Agents

Beaten on benchmarks

Head-to-head results where a newer method reports beating MemoryBank. Values are copied from the source paper's tables — verify against the cited paper.

What to use instead

Recent methods in the same sub-problem, not yet superseded in the knowledge base — arXiv benchmark leaders, not vetted production recommendations.