引入具有时间推理功能的更新版Token-Efficient Memory算法
TL;DR · AI 摘要
mem0发布更新的Token-Efficient Memory算法，提升时间推理和多会话推理性能，分别提高3.8分和1.5分。
核心要点
- 时间推理性能提升3.8分
- 多会话推理性能提升1.5分
- 算法保持每调用检索低于7,000 tokens
结构提纲
按章节快速跳转。
思维导图
用一张图看清主题之间的关系。
查看大纲文本（无障碍 / 无 JS 友好）
- Token-Efficient Memory算法更新
- 核心改进
- 时间推理+3.8分
- 多会话推理+1.5分
- 性能表现
- LoCoMo 92.5分
- LongMemEval 94.4分
- 效率控制
- 每调用检索<7,000 tokens
金句 / Highlights
值得收藏与分享的关键句。
时间推理性能提升3.8分，多会话推理性能提升1.5分。
算法在LoCoMo和LongMemEval测试中分别达到92.5和94.4分。
每调用检索保持在7,000 tokens以下。
During our April algorithm launch, we called out multi-session reasoning and temporal reasoning as two of the hardest categories in memory.
This release improves both: +3.8 pts on temporal https://t.co/lXJt0521E8" / X
mem0 on X: "Introducing the updated Token-Efficient Memory algorithm with Temporal Reasoning During our April algorithm launch, we called out multi-session reasoning and temporal reasoning as two of the hardest categories in memory. This release improves both:+3.8 pts on temporal https://t.co/lXJt0521E8" / X
Don’t miss what’s happening
Introducing the updated Token-Efficient Memory algorithm with Temporal Reasoning During our April algorithm launch, we called out multi-session reasoning and temporal reasoning as two of the hardest categories in memory. This release improves both:+3.8 pts on temporal reasoning and+1.5 pts on multi-session at top_200 with new features: Temporal Reasoning and Memory Decay Mem0 now reaches 92.5 on LoCoMo and 94.4 on LongMemEval, while keeping retrieval under<7,000 tokens per call.
·
8
13
58
30
Read 8 replies