Prompt-Engineering-Guide/pages/techniques/activeprompt.zh.mdx
RealTong 3de62e5bd3
Update: Chinese translation
- Update Chinese translation of the techniques section
- For a better reading experience on the Chinese internet, modify the spacing between Chinese and English characters.
2024-03-02 18:47:57 +08:00

12 lines
1004 B
Plaintext
Raw Blame History

This file contains ambiguous Unicode characters

This file contains Unicode characters that might be confused with other characters. If you think that this is intentional, you can safely ignore this warning. Use the Escape button to reveal them.

# Active-Prompt
import { Callout, FileTree } from 'nextra-theme-docs'
import {Screenshot} from 'components/screenshot'
import ACTIVE from '../../img/active-prompt.png'
思维链CoT方法依赖于一组固定的人工注释范例。问题在于这些范例可能不是不同任务的最有效示例。为了解决这个问题[Diao 等人2023](https://arxiv.org/pdf/2302.12246.pdf)最近提出了一种新的提示方法,称为 Active-Prompt以适应 LLMs 到不同的任务特定示例提示(用人类设计的 CoT 推理进行注释)。
下面是该方法的说明。第一步是使用或不使用少量 CoT 示例查询 LLM。对一组训练问题生成 *k* 个可能的答案。基于 *k* 个答案计算不确定度度量(使用不一致性)。选择最不确定的问题由人类进行注释。然后使用新的注释范例来推断每个问题。
<Screenshot src={ACTIVE} alt="ACTIVE" />
图片来源:[Diao等人2023](https://arxiv.org/pdf/2302.12246.pdf)