AI · NLP · ML · Engineering Research home · systemized notes · engineering archive

Practical AI Notes, Systems, and Engineering

围绕 NLP、LLM、机器学习与工程实践，持续沉淀值得复用的方法、系统设计、实验记录与项目思考。

Explore Posts Know More
Focus NLP / LLM / ML
Style Notes · Practice · Systems
Content Articles · Experiments · Engineering
Archive 174 posts indexed
Signal PAPER highlighted
Mode Research / Build / Review
Dashboard Surface
Research Feed
Latest notes · experiments · engineering logs
Featured Note ▣PAPER ACTIVE

SELFDOUBT：给推理型 LLM 加一个低成本、可部署的不确定性后验层

SELFDOUBT：给推理型 LLM 加一个低成本、可部署的不确定性后验层 论文：Uncertainty Quantification for Reasoning LLMs via the Hedge-to-Verify Ratio 链接：https://arxiv.org/abs/2604.06389 说明：这篇笔记主要依据 arXiv 摘要与 HTML 正文完成；我没有假装已经逐页复核全部附录与代码实现。 ...
2026-04-11 · 记录 · LLM · Reasoning · Uncertainty · Calibration · AI Agent
Research Timeline
Stream Log
Recent entries from notes, project logs, papers, and implementation traces.
FEATURED TAGS
数学 llm agent github llms 工具 paper workflow mcp rag 开发 推荐系统 机器学习 语音识别 developer-tools multi-agent nlp 搜索 深度学习 memory tool-use arxiv coding-agent daily-pick reinforcement-learning spark AI Agent bert deer-flow planning retrieval sandbox skills verification 信息抽取 开发语言 数据结构 LLM Multi-Agent code-intelligence coordination gui-agent hallucination hierarchy langchain multimodal observability orchestration orgagent project safety token-efficiency
资源链接

记录