Expired95% confidenceEventTime unknown
Researchers from KAST University of Tübingen and Never AI Lab proposed the MEME (Multi-Entity Evolving Memory Evaluation) benchmark for systematically evaluating the performance of LLM memory systems on dependency reasoning.
1
Sources
95%
Confidence
Medium-term (~90 days)
Relevance
6/1/2026
First Seen
Valid until: 8/30/2026(expired)
Sources
Related Entities
Related Claims
UnverifiedThe memory stream paradigm of Stanford Generative Agents retrieves memories through three dimensions—Recency, Importance, and Relevance—where Importance is self-rated by the model on a scale of 1–10 and Relevance is calculated via embedding vector cosine similarity.71% similarUnverifiedThe memory of LLMs depends on the context window, which is the maximum number of tokens the model can process in a single inference.69% similarUnverifiedMIT的方法将检测目标从输出层转移到模型内部特征层面，通过分析权重、激活模式或表征空间的统计特征推断模型是否形成记忆68% similarUnverifiedMemory机制在大语言模型工程中对应系统提示词工程与外部记忆存储的结合，用于模拟无状态模型的持久化状态67% similarUnverifiedSemantic agents with LLMs at their core understand task intent rather than memorizing operation paths, theoretically enabling a significant improvement in cross-version robustness.67% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/21541API
curl https://kongchang.com/api/v1/knowledge/claims/21541MCP
get_claim(id=21541)