Feed Entities Podcasts Deep Dives Developers About

[KongchangAI]

Real-time AI knowledge base. Every claim is traceable to its source, with data freshness under 24 hours.

Powered by 240+ first-hand AI sources

Knowledge

Feed
Hot Entities
Podcasts

Developers

CLI
MCP Server
REST API
About

© 2026 KongchangAI. The freshest, most accurate AI knowledge

Unverified80% confidenceOpinionExact time

Locomo评测数据集存在局限性，不同Memory框架使用不同的judgment prompt导致分数缺乏可比性

1

Sources

80%

Confidence

Long-term

Relevance

5/16/2026

First Seen

Sources

memU记忆框架解读：用文件系统统一多模态Agent记忆

bilibili日新月异max5/16/2026

Related Entities

Related Claims

Verified若用含测试集的全量数据计算均值和标准差，会导致预处理泄漏，使评估指标虚高，无法真实反映模型泛化能力72% similar Unverified该量表属于自评量表，受被试主观报告倾向影响大；存在掩饰动机或述情障碍的人群测评准确性会明显下降72% similar Unverified误差累积问题往往源于参照系选择而非单纯模型容量限制72% similar Unverified语义缓存阈值设定是一个权衡：阈值过高导致命中率低，过低则可能将语义不同的问题错误映射引发响应质量问题71% similar Unverified限额校验往往发生在请求入队时，而实际token消耗统计发生在推理完成后，两者之间存在时间差70% similar

Cite This Claim

Stable URI

https://kongchang.com/claim/1568

API

curl https://kongchang.com/api/v1/knowledge/claims/1568

MCP

get_claim(id=1568)