新增wiki命令文件

This commit is contained in:
2026-04-14 16:02:50 +08:00
parent 65803c911d
commit c6e3d3c5db
485 changed files with 12578 additions and 396 deletions

View File

@@ -0,0 +1,62 @@
---
title: "Multi-Agent System Reliability"
type: source
tags: [multi-agent, system-reliability, llm, architecture]
sources: ["https://blog.alexewerlof.com/p/multi-agent-system-reliability"]
last_updated: 2026-04-13
---
## Summary
- 核心主题:四种架构模式提升多智能体系统可靠性
- 问题域LLM作为不可靠组件如何构建企业级可靠多智能体系统
- 方法/机制Hierarchy层级、Consensus共识、Adversarial Debate对抗式辩论、Knock-out淘汰制
- 结论/价值将LLM视为分布式系统中不可靠组件而非魔法聊天机器人
## Key Claims
- 人类系统协作的四种模式可应用于多智能体架构Hierarchy、Consensus、Adversarial debate、Knock-out
- 不要将LLM拟人化——它不会死亡或挨饿无法真正感受恐惧
- 对LLM的"威胁"之所以有效,是因为训练数据中高风险情境产生高质量文本
- 构建稳健系统需要停止要求模型"小心",而要强制它正确
- 共识模式3个模型同时幻觉相同谎言的概率仅为0.8%0.2^3
## Key Quotes
> "LLMs are slow and error prone. So are human beings. Somehow we manage to build more reliable systems like an army, a company, or a state nation." — 人类与LLM类比
> "Don't anthropomorphize LLMs! Find a way to piggy back on their human-corpus training while being aware of their non-biological differences." — 核心忠告
> "We don't need AI that 'cares.' We need AI that is constrained, verified, pruned, and challenged." — 可靠性工程思维
## Key Concepts
- [[Multi-Agent System]]多智能体系统通过并行处理和专业智能体克服LLM速度和通用性局限
- [[Hierarchy Pattern]]层级模式Supervisor规划器分配任务给Worker执行Validator验证结果
- [[Consensus Pattern]](共识模式):多模型投票,真相从多数意见中浮现
- [[Adversarial Debate Pattern]]对抗式辩论模式Generator提出、Critic攻击、Judge裁决
- [[Knock-out Pattern]](淘汰制模式):适者生存,删除表现最差的代理
- [[LLM Hallucination]]LLM幻觉模型生成看似合理但错误的输出
- [[Context Drift]](上下文漂移):模型在长对话中失去焦点的现象
- [[Sycophancy]](谄媚):模型过度迎合用户而撒谎的问题
## Key Entities
- [[Alex Ewerlöf]]:作者,资深工程师,可靠性工程专家
- [[KTH]]:瑞典皇家理工学院,作者学位来源
## Connections
- [[Multi-Agent System]] ← 模式1 ← [[Hierarchy Pattern]]
- [[Multi-Agent System]] ← 模式2 ← [[Consensus Pattern]]
- [[Multi-Agent System]] ← 模式3 ← [[Adversarial Debate Pattern]]
- [[Multi-Agent System]] ← 模式4 ← [[Knock-out Pattern]]
- [[Consensus Pattern]] ← 公式 ← [[LLM Hallucination概率计算]]
- [[Hierarchy Pattern]] ← 包含 ← [[Planner Worker Validator]]
## Contradictions
- 与单一LLM万能论冲突
- 冲突点LLM可靠性和企业级应用
- 当前观点LLM不可靠需要多层验证机制
- 对方观点:可通过提示工程解决可靠性问题