六层技术版图
从 L0 Financial Agents（最内层）到 L5 Algorithmic Foundations（最外层）的同心结构 —— 每一层向外是它的技术基座。点击子层查看论文。
L0 · Financial Agents金融智能体
25 篇 →最内层 · 任何用 LLM/agent 范式处理金融 task 的系统:客服、交易决策、alpha 挖掘、quant R&D、风控
L0.1 Customer-Facing Financial Agents
L0.2 Trading-Decision Agents
L0.3 Alpha Mining / Factor Discovery（主战场）
L0.4 Quant R&D Agents
L0.5 Risk / Safety / Reproducibility
L1 · Domain Research Agents领域研究智能体
18 篇 →LLM agent 自主完成 domain-specific 研究:hypothesis → experiment → analysis
FunSearch · 2023AI Scientist v1 · 2024AI Scientist v2 · 2025ASI-ARCH · 2025AlphaProof / AlphaGeometry 2 · 2024AlphaEvolve · 2025Coscientist · 2023ChemCrow · 2023BioPlanner · 2024GNoME · 2023From Copilots to Colleagues · 2026Deep Research · 2025Deep Research Agents · 2025Deep Research · 2025Reinforcement Learning Foundations for D… · 2025Agentic AI for Scientific Discovery · 2025Highly Accurate Protein Structure Predic… · 2021Discovering Faster Matrix Multiplication… · 2022
L2 · Production Agent Systems生产级智能体系统
11 篇 →真正部署在生产环境的 agent:coding、computer use、deep research、business automation
L3 · LLM Agent Patterns & Frameworks智能体模式与框架
34 篇 →不依赖 domain 的 agent 通用 building block:reasoning、planning、tool、memory、multi-agent
L3.1 Reasoning Patterns
Chain-of-Thought Prompting Elicits Reaso… · 2022Self-Consistency Improves Chain of Thoug… · 2022Tree of Thoughts · 2023Graph of Thoughts · 2023LATS · 2023Reflexion · 2023Self-Refine · 2023PAL · 2022LLM-based Agentic Reasoning Frameworks · 2025A Survey of Frontiers in LLM Reasoning · 2025Logical Reasoning in Large Language Models · 2025
L3.2 Planning Patterns
L3.3 Tool Use & Integration
L3.4 Memory Architectures
L4 · Foundation Model Tech Stack基础模型技术栈
14 篇 →foundation model 的训练、推理、扩展技术,每 3-6 个月翻一轮
L4.1 Pretraining / Architecture
Attention is All You Need · 2017Scaling Laws for Neural Language Models · 2020Training Compute-Optimal Large Language … · 2022DeepSeek-V3 Technical Report · 2024DeepSeek-V4 Technical Report · 2026Mamba · 2023Encoding Recurrence into Transformers · 2023On the Opportunities and Risks of Founda… · 2021
L4.2 Post-Training (Alignment)
L4.3 Test-Time Compute
L5 · Algorithmic Foundations算法基础
12 篇 →最外层 · 不依赖 LLM 的经典 AI/ML 算法,agent 范式的真正地基
L5.1 Reinforcement Learning
Q-Learning · 1989Playing Atari with Deep Reinforcement Le… · 2013Mastering the Game of Go with Deep Neura… · 2016Mastering Chess and Shogi by Self-Play (… · 2017Mastering Atari, Go, Chess and Shogi by … · 2019Proximal Policy Optimization (PPO) · 2017Decision Transformer · 2021The Landscape of Agentic Reinforcement L… · 2025
L5.2 Search & Planning
L5.3 Probabilistic / Conformal