547334452fuhai/Prompts-that-Can-Suppress-AI-Hallucinations

GitHub: 547334452fuhai/Prompts-that-Can-Suppress-AI-Hallucinations

一套融合多学科经典方法的AI提示词模板,旨在通过结构化的结果校验与过程质量控制机制来有效抑制大语言模型的幻觉问题。

Stars: 3 | Forks: 0

Prompts that Can Suppress AI Hallucinations Prompts That Can Reduce AI Hallucinations 可以抑制AI幻觉的提示词 可以减少AI幻觉的提示词 Purpose: (1) Answering questions related to daily life and work. (2) Conducting specialized research in specific fields. (3) Preliminary research for technical solutions and problem-solving strategies. (4) Preliminary research for innovative problems and tasks. (5) Serving as a foundational platform, tool, and engine for generating more professional-grade and productivity-enhancing prompts. 用途: (1)日常生活和工作的问题回答。 (2)垂直领域的专业调研。 (3)技术方案/问题解决方案的前置调研。 (4)创新问题/任务的前置调研。 (5)生成更多的专业级/生产力级提示词的生产底座、工具与母机。 Features: (1) Includes result verification, and the basis for conclusions can be manually verified by clicking on links. (2) Includes process verification and quality control, using professional methods. (3) Reduces AI hallucinations and minimizes the manual verification workload for critical tasks. (4) Reduces AI hallucinations and minimizes the introduction of “thought viruses” caused by AI hallucinations. (5) Reduces AI hallucinations and minimizes serious losses caused by AI hallucinations. 特点: (1)带结果校验,并且,结论的依据可人工点击链接核查。 (2)带过程校验与质量控制,采用专业方法。 (3)减少AI幻觉,减少重要任务中人工校核工作量。 (4)减少AI幻觉,减少因AI幻觉导致的思维病毒注入。 (5)减少AI幻觉,降低因AI幻觉导致的严重损失。 Principle: This approach deeply integrates classic methods such as open-source intelligence, logical analysis, quality control of processes and results, cybernetics and information theory, and management engineering. For specific details, please refer to the content in the prompt template. 原理: 深度融合开源情报学、逻辑分析、过程和结果的质量控制、控制论与信息论、管理工程等经典方法,具体内容可参考提示词模板内容。 Instructions: (1) In the sections of the prompt template where content is required, enter a description of your task or question. We recommend that you organize your description as clearly as possible. (2) Each section of the prompt template can be extensively modified, expanded, or deleted according to your specific needs or context. 使用说明: (1)在提示词模板里需要输入内容的地方,输入您的任务/问题描述,建议尽量条理清晰。 (2)在提示词模板里每个环节均可以根据具体需求/场景,进行深度修改、增加、删除。 Proposed and developed by Hai Fu, https://orcid.org/0000-0002-6322-3186. Email: fuhai19890909@163.com MIT Open Source License. Copyright © [Hai Fu/Fu Hai/ORCID (0000-0002-6322-3186)] Anyone may freely access, copy, modify, integrate, and distribute this prompt document. It supports all scenarios, including personal use, commercial use, closed-source secondary development, and paid commercial use, without the need for prior notification or payment of licensing fees. All publicly distributed, commercially used, or republished copies must retain this original copyright notice and this license text; the original attribution must not be deleted or altered. Proposed and developed by Hai Fu, https://orcid.org/0000-0002-6322-3186. Email: fuhai19890909@163.com MIT 开源许可 版权所有 © [Hai Fu/符海/orcid(0000-0002-6322-3186)] 任何人可免费获取、复制、修改、整合、分发、本提示词文档,支持个人使用、商业使用、闭源二次开发、付费商用等全部场景,无需提前告知、无需支付授权费用。 所有公开分发、商用、二次发布的副本,必须保留本原始版权声明与本许可文本,不得删除、篡改原创署名。
标签:AI幻觉抑制, DLL 劫持, 大语言模型, 提示词工程, 策略决策点, 结果校验, 质量控制, 防御加固