Unverified50% confidenceBenchmarkExact time
长程记忆评测集（如 LongMemEval、MemGPT 基准）考察五个维度：单跳信息提取、跨会话推理、知识更新、时间推理、拒答判断
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
9/22/2026
First Seen
Valid until: 12/21/2026
Sources
Related Entities
Related Claims
UnverifiedMemGAS在四个长期记忆基准上进行评估：Locomo、LongMemEval-S、LongMemEval-M和LongMTBench Plus76% similarUnverifiedLoCoMo和LongMemEval等对话回忆基准测量的是对对话历史的问答能力，而非记忆是否改变了工具使用智能体的实际行为75% similarUnverified记忆冲突解决涉及时间戳优先级与置信度评分的复合策略74% similarUnverified记忆增强型智能体通过持续读取用户历史交互数据来识别高频关注点，技术实现通常需对原始对话数据做向量化处理再通过相似度检索或关键词抽取筛选73% similarUnverified生成式智能体的记忆检索基于重要性、近期性和相关性评分提取关键记忆73% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/941823API
curl https://kongchang.com/api/v1/knowledge/claims/941823MCP
get_claim(id=941823)