学术相关
Sequential learning under informational ambiguity 1
Is Competition Only One Click Away? The Digital Markets Act’s Impact on Google Maps 2
技术技巧
DeepHub IMBA丨多智能体模式
单智能体在功能复杂情况下会变得不可用，该临界点被称为“单智能体墙” （Single-agent capacity threshold）
Once intelligence reaches a threshold, multi-agent systems become a vital way to scale performance. … Our internal evaluations show that multi-agent research systems excel especially for breadth-first queries that involve pursuing multiple independent directions simultaneously. We found that a multi-agent system with Claude Opus 4 as the lead agent and Claude Sonnet 4 subagents outperformed single-agent Claude Opus 4 by 90.2% on our internal research eval.3
核心原因是指令迷雾和工具过载
常见的多智能体模式
- Sequential Muti-Agent
- Parallel Multi-Agent
- Layered or Hierarchical
- Router
- 与layer的差异是不分解任务，而是直接分发
- Reflect and Critique
- Custom and Consensus
-
Chen, J. Y. (2026). Sequential learning under informational ambiguity. American Economic Review, 116(1), 209–245. https://doi.org/10.1257/aer.20231394 ↩
-
Pape, L.-D., & Rossi, M. (2026). Is Competition Only One Click Away? The Digital Markets Act’s Impact on Google Maps. Marketing Science, mksc.2025.159. https://doi.org/10.1287/mksc.2025.0159 ↩
-
Engineering at Anthropic. (2025, June 13). How we built our multi-agent research system. https://www.anthropic.com/engineering/multi-agent-research-system ↩