---
toAstro: true
astroType: null
published: true
toWexin: null
toJuejin: null
toZhihu: null
title: ChainForge简单介绍
date: 2024-05-22T00:00:00.000Z
author: KazooTTT
tags:
  - 工具
  - chainforge
  - llm
finished: true
category: AI
slug: chainforge-intro
description: >-
  ChainForge is a visual programming environment designed for prompt
  engineering, allowing users to evaluate and test prompts for large language
  models (LLMs). It supports various use cases such as assessing prompt
  effectiveness, conducting adversarial testing, and evaluating model
  performance. The platform also facilitates the customization of LLMs through
  its documentation on adding custom providers. Users can find practical
  examples and tutorials on the ChainForge website and GitHub repository,
  including a case study on Twitter and a video demonstrating LLM prompt
  injection attacks and vulnerability testing. Additionally, ChainForge is
  working on enhancing its functionality by enabling internal and proxy network
  environments.
NotionID-notionnext: 40ec4f8d-2030-4ce1-b8c7-c1c9f56ef55b
link-notionnext: 'https://kazoottt.notion.site/ChainForge-40ec4f8d20304ce1b8c7c1c9f56ef55b'
date_created: 20250104
date_modified: 20250304
---

# ChainForge 简单介绍

![Pasted image 20240417222811](<https://pictures.kazoottt.top/2024/05/20240522-c31876b89dcb822e411af1ef7992eb49.png>)

官网：

[ChainForge: A visual programming environment for prompt engineering](<https://chainforge.ai/>)

github 地址：

[GitHub - ianarawjo/ChainForge: An open-source visual programming environment for battle-testing prompts to LLMs.](<https://github.com/ianarawjo/ChainForge>)

## 简介

![Pasted image 20240417223400](<https://pictures.kazoottt.top/2024/05/20240522-a9e96bff073602718c11943e61bcf758.png>)

## 使用场景

评估提示词

攻击性测试

评估模型

评估某项指标

## 实践

![Pasted image 20240417231517](<https://pictures.kazoottt.top/2024/05/20240522-5fea389ac9bca4a43c076dc442d374e6.png>)

这两个例子单独拿出来讲

## TODO

1. 内部环境运行
2. 内网环境 proxy [Add Custom Providers - ChainForge Documentation](<https://chainforge.ai/docs/custom_providers/> 提供了自定义 llm

## 参考

使用案例：

[X](<https://twitter.com/9hills/status/1772467763165958602>)

自动化 Prompts

[\[2402.10949\] The Unreasonable Effectiveness of Eccentric Automatic Prompts](<https://arxiv.org/abs/2402.10949>)

相关的视频

[LLM Prompt Injection Attacks & Testing Vulnerabilities With ChainForge - YouTube](<https://www.youtube.com/watch?v=x1_uEnoxvNs>)

使用 ChainForge 进行 LLM 提示注入攻击和漏洞测试