-
Notifications
You must be signed in to change notification settings - Fork 7.9k
Add and Enhance Chinese Translation of Techniques Section #729
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
base: main
Are you sure you want to change the base?
Changes from all commits
e1ea3f5
b5b0c9a
88711d7
b721714
9038f30
5dbde89
3be928d
ae1bf59
451c8d4
File filter
Filter by extension
Conversations
Jump to
Diff view
Diff view
There are no files selected for viewing
| Original file line number | Diff line number | Diff line change |
|---|---|---|
| @@ -0,0 +1,73 @@ | ||
| # 元提示词(Meta Prompting) | ||
| import { CoursePromo, CoursesSection, CourseCard } from '../../components/CourseCard' | ||
|
|
||
| ## 介绍 | ||
|
|
||
| 元提示词是一种高级提示技术,它关注任务和问题的结构与句法层面,而非具体的细节内容。元提示词的目标是构建一种与大语言模型(LLM)交互的更抽象、结构化的方式,强调信息的形式与模式,而非传统的以内容为中心的方法。 | ||
|
|
||
| ## 核心特性 | ||
|
|
||
| 根据[Zhang等人,2024](https://arxiv.org/abs/2311.11482)的研究,元提示词的核心特性可归纳如下: | ||
|
|
||
| **1. 结构导向**: 优先考虑问题和解决方案的格式与模式,而非具体内容。 | ||
|
|
||
| **2. 句法聚焦**: 利用句法作为预期回答或解决方案的引导模板。 | ||
|
|
||
| **3. 抽象示例**: 使用抽象示例作为框架,展示问题和解决方案的结构,而不关注具体细节。 | ||
|
|
||
| **4. 多用途**: 适用于各种领域,能够针对广泛的问题提供结构化的响应。 | ||
|
|
||
| **5. 分类方法**: 借助类型理论,强调提示中组件的分类和逻辑排列。 | ||
|
|
||
| ## 与少样本提示 (Few-Shot Prompting) 的优势对比 | ||
|
|
||
| [Zhang等人, 2024](https://arxiv.org/abs/2311.11482) 指出,元提示词与少样本提示的不同之处在于,元提示词侧重于结构导向的方法,而少样本提示则强调内容驱动的方法。 | ||
|
|
||
| 以下示例(原自[Zhang等人, 2024](https://arxiv.org/abs/2311.11482))展示了在解决 MATH 基准测试问题时,结构化元提示词与少样本提示词之间的区别: | ||
|
|
||
|  | ||
|
|
||
| 元提示词相比少样本提示的优势包括: | ||
|
|
||
| **1. Token 效率**: 通过聚焦结构而非详细内容,减少了所需的 Token 数量。 | ||
|
|
||
| **2. 公平比较**: 通过最小化具体示例的影响,为比较不同解题模型提供了一种更公平的方法。 | ||
|
|
||
| **3. 零样本效能**: 可以被视为一种特殊的零样本提示,使具体示例的干扰降至最低。 | ||
|
|
||
| ## 应用场景 | ||
|
|
||
| 通过聚焦解题的结构模式,元提示词为处理复杂话题提供了清晰的路线图,增强了 LLM 在各个领域的推理能力。 | ||
|
|
||
| 需要注意的是,元提示词假设 LLM 对所处理的特定任务或问题具有内在知识。虽然 LLM 可以泛化到未见过的任务,但对于非常独特且新颖的任务,元提示词的表现可能会像零样本提示一样有所下降。 | ||
|
|
||
| 元提示词受益的应用领域包括但不限于:复杂推理任务、数学问题求解、编程挑战以及理论查询。 | ||
|
|
||
| <CoursesSection title="Related Learning"> | ||
| <CourseCard | ||
| tag="Course" | ||
| tagColor="blue" | ||
| title="Prompt Engineering for LLMs" | ||
| description="Master meta prompting, structure-oriented techniques, and advanced methods for complex reasoning." | ||
| href="https://academy.dair.ai/courses/introduction-prompt-engineering" | ||
| level="Beginner" | ||
| duration="2 hours" | ||
| /> | ||
| <CourseCard | ||
| tag="Course" | ||
| tagColor="purple" | ||
| title="Building Effective AI Agents" | ||
| description="Learn to build effective AI agents. Covers function calling, tool integration, and debugging agentic systems." | ||
| href="https://academy.dair.ai/courses/building-effective-ai-agents" | ||
| level="Intermediate" | ||
| duration="5 hours" | ||
| /> | ||
| </CoursesSection> | ||
|
|
||
| <CoursePromo | ||
| title="Explore All Courses" | ||
| description="Discover our full catalog of AI and prompt engineering courses. From beginners to advanced practitioners." | ||
| href="https://academy.dair.ai/" | ||
| buttonText="Browse Academy" | ||
| promoCode="PROMPTING20" | ||
| /> |
| Original file line number | Diff line number | Diff line change |
|---|---|---|
| @@ -1,6 +1,6 @@ | ||
| # 自我反思(Reflexion) | ||
|
|
||
| 自我反思是一个通过语言反馈来强化基于语言的智能体的框架。根据 [Shinn et al. (2023)](https://arxiv.org/pdf/2303.11366.pdf),“自我反思是一种‘口头’强化的新范例,它将策略参数化为智能体的记忆编码与 LLM 的参数选择配对。” | ||
| 自我反思是一个通过语言反馈来强化基于语言的智能体的框架。根据 [Shinn等人, (2023)](https://arxiv.org/pdf/2303.11366.pdf),“自我反思是一种‘口头’强化的新范例,它将策略参数化为智能体的记忆编码与 LLM 的参数选择配对。” | ||
|
||
|
|
||
| 在高层次上,自我反思将来自环境的反馈(自由形式的语言或者标量)转换为语言反馈,也被称作 **self-reflection**,为下一轮中 LLM 智能体提供上下文。这有助于智能体快速有效地从之前的错误中学习,进而提升许多高级任务的性能。 | ||
|
|
||
|
|
||
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
The ordering of "meta-prompting" is inconsistent with the English version. In the English _meta.en.json, "meta-prompting" appears on line 5 (after "cot" and before "consistency"), but in the Chinese version it's placed at line 19 (at the end of the list). This ordering difference may affect the navigation menu structure and user experience. Consider aligning the order with the English version by placing "meta-prompting" between "cot" and "consistency" for consistency across language versions.