ChatGPT/FastGPT
1
0
Fork 0
mirror of https://github.com/labring/FastGPT.git synced 2025-07-29 09:44:47 +00:00
Code Issues Packages Projects Releases Wiki Activity

4.8.13 feature (#3118)

Browse Source 
* chore(ui): login page & workflow page (#3046)

* login page & number input & multirow select & llm select

* workflow

* adjust nodes

* New file upload (#3058)

* feat: toolNode aiNode readFileNode adapt new version

* update docker-compose

* update tip

* feat: adapt new file version

* perf: file input

* fix: ts

* feat: add chat history time label (#3024)

* feat:add chat and logs time

* feat: add chat history time label

* code perf

* code perf

---------

Co-authored-by: 勤劳上班的卑微小张 <jiazhan.zhang@ggimage.com>

* add chatType (#3060)

* pref: slow query of full text search (#3044)

* Adapt findLast api;perf: markdown zh format. (#3066)

* perf: context code

* fix: adapt findLast api

* perf: commercial plugin run error

* perf: markdown zh format

* perf: dockerfile proxy (#3067)

* fix ui (#3065)

* fix ui

* fix

* feat: support array reference multi-select (#3041)

* feat: support array reference multi-select

* fix build

* fix

* fix loop multi-select

* adjust condition

* fix get value

* array and non-array conversion

* fix plugin input

* merge func

* feat: iframe code block；perf: workflow selector type (#3076)

* feat: iframe code block

* perf: workflow selector type

* node pluginoutput check (#3074)

* feat: View will move when workflow check error;fix: ui refresh error when continuous file upload (#3077)

* fix: plugin output check

* fix: ui refresh error when continuous file upload

* feat: View will move when workflow check error

* add dispatch try catch (#3075)

* perf: workflow context split (#3083)

* perf: workflow context split

* perf: context

* 4.8.13 test (#3085)

* perf: workflow node ui

* chat iframe url

* feat: support sub route config (#3071)

* feat: support sub route config

* dockerfile

* fix upload

* delete unused code

* 4.8.13 test (#3087)

* fix: image expired

* fix: datacard navbar ui

* perf: build action

* fix: workflow file upload refresh (#3088)

* fix: http tool response (#3097)

* loop node dynamic height (#3092)

* loop node dynamic height

* fix

* fix

* feat: support push chat log (#3093)

* feat: custom uid/metadata

* to: custom info

* fix: chat push latest

* feat: add chat log envs

* refactor: move timer to pushChatLog

* fix: using precise log

---------

Co-authored-by: Finley Ge <m13203533462@163.com>

* 4.8.13 test (#3098)

* perf: loop node refresh

* rename context

* comment

* fix: ts

* perf: push chat log

* array reference check & node ui (#3100)

* feat: loop start add index (#3101)

* feat: loop start add index

* update doc

* 4.8.13 test (#3102)

* fix: loop index;edge parent check

* perf: reference invalid check

* fix: ts

* fix: plugin select files and ai response check (#3104)

* fix: plugin select files and ai response check

* perf: text editor selector;tool call tip;remove invalid image url;

* perf: select file

* perf: drop files

* feat: source id prefix env (#3103)

* 4.8.13 test (#3106)

* perf: select file

* perf: drop files

* perf: env template

* 4.8.13 test (#3107)

* perf: select file

* perf: drop files

* fix: imple mode adapt files

* perf: push chat log (#3109)

* fix: share page load title error (#3111)

* 4.8.13 perf (#3112)

* fix: share page load title error

* update file input doc

* perf: auto add file urls

* perf: auto ser loop node offset height

* 4.8.13 test (#3117)

* perf: plugin

* updat eaction

* feat: add more share config (#3120)

* feat: add more share config

* add i18n en

* fix: missing subroute (#3121)

* perf: outlink config (#3128)

* update action

* perf: outlink config

* fix: ts (#3129)

* 更新 docSite 文档内容 (#3131)

* fix: null pointer (#3130)

* fix: null pointer

* perf: not input text

* update doc url

* perf: outlink default value (#3134)

* update doc (#3136)

* 4.8.13 test (#3137)

* update doc

* perf: completions chat api

* Restore docSite content based on upstream/4.8.13-dev (#3138)

* Restore docSite content based on upstream/4.8.13-dev

* 4813.md缺少更正

* update doc (#3141)

---------

Co-authored-by: heheer <heheer@sealos.io>
Co-authored-by: papapatrick <109422393+Patrickill@users.noreply.github.com>
Co-authored-by: 勤劳上班的卑微小张 <jiazhan.zhang@ggimage.com>
Co-authored-by: Finley Ge <32237950+FinleyGe@users.noreply.github.com>
Co-authored-by: a.e. <49438478+I-Info@users.noreply.github.com>
Co-authored-by: Finley Ge <m13203533462@163.com>
Co-authored-by: Jiangween <145003935+Jiangween@users.noreply.github.com>
This commit is contained in:
Archer
2024-11-13 11:29:53 +08:00
committed by shilin66
parent 72777c341b
commit f680be52f1
449 changed files with 7626 additions and 4180 deletions
Download Patch File Download Diff File Expand all files Collapse all files

8
docSite/content/zh-cn/docs/use-cases/app-cases/_index.md Normal file
View File

@@ -0,0 +1,8 @@
---
title: "应用搭建案例"
description: "FastGPT 应用场景及功能实现的搭建案例"
icon: "construction"
draft: false
weight: 600
---
<!-- 600 ~ 700 -->

471
docSite/content/zh-cn/docs/use-cases/app-cases/dalle3.md Normal file
View File

@@ -0,0 +1,471 @@
---
title: 'Dalle3 绘图'
description: '使用 HTTP 模块绘制图片'
icon: 'image'
draft: false
toc: true
weight: 614
---
| | |
| --------------------- | --------------------- |
| ![](/imgs/demo-dalle1.webp) | ![](/imgs/demo-dalle2.webp) |
## OpenAI Dalle3 接口
先来看下官方接口的参数和响应值：
Body
```json
{
"model": "dall-e-3",
"prompt": "A cute baby sea otter",
"n": 1,
"size": "1024x1024"
}
```
Response
```json
{
"created": 1589478378,
"data": [
{
"url": "https://..."
},
{
"url": "https://..."
}
]
}
```
## 编排思路
1. 通过 AI 来优化图片绘制的提示词（这步省略了，自己找提示词即可）
2. 通过 `【HTTP 请求】模块` 调用 Dalle3 接口，获取图片的 URL。
3. 通过 `【文本加工】模块` 来构建 `Markdown` 的图片格式。
4. 通过 `【指定回复】模块` 来直接输出图片链接。
### 1. 构建 HTTP 模块
请求参数直接复制 Dalle3 接口的即可，并求改 prompt 为变量。需要增加一个 `Headers.Authorization` 。
Body:
```json
{
"model": "dall-e-3",
"prompt": "{{prompt}}",
"n": 1,
"size": "1024x1024"
}
```
Headers:
`Authorization: Bearer sk-xxx`
Response:
响应值需要根据 Dalle3 接口的返回值进行获取，我们只绘制了1张图片，所以只需要取第一张图片的 URL 即可。给 HTTP 模块增加一个自定义输出 `data[0].url` 。
### 2. 文本加工 - 构建图片链接
在 `Markdown` 语法中 `![图片描述](图片链接)` 表示插入图片，图片链接由【HTTP 请求】模块输出。
因此可以增加一个输入来接收 `【HTTP 请求】模块` 的图片链接输出，并在 `【文本加工】模块 - 文本` 中通过变量来引用图片链接，从而得到一个完整的 `Markdown` 图片格式。
### 3. 指定回复
指定回复可以直接输出传入的内容到客户端，因此可以直接输出加工好的 `Markdown` 图片格式即可。
## 编排代码
{{% details title="编排配置" closed="true" %}}
```json
{
"nodes": [
{
"nodeId": "userGuide",
"name": "系统配置",
"intro": "可以配置应用的系统参数",
"avatar": "/imgs/workflow/userGuide.png",
"flowNodeType": "userGuide",
"position": {
"x": 531.2422736065552,
"y": -486.7611729549753
},
"inputs": [
{
"key": "welcomeText",
"renderTypeList": [
"hidden"
],
"valueType": "string",
"label": "core.app.Welcome Text",
"value": ""
},
{
"key": "variables",
"renderTypeList": [
"hidden"
],
"valueType": "any",
"label": "core.app.Chat Variable",
"value": []
},
{
"key": "questionGuide",
"valueType": "boolean",
"renderTypeList": [
"hidden"
],
"label": "core.app.Question Guide",
"value": false
},
{
"key": "tts",
"renderTypeList": [
"hidden"
],
"valueType": "any",
"label": "",
"value": {
"type": "web"
}
},
{
"key": "whisper",
"renderTypeList": [
"hidden"
],
"valueType": "any",
"label": "",
"value": {
"open": false,
"autoSend": false,
"autoTTSResponse": false
}
},
{
"key": "scheduleTrigger",
"renderTypeList": [
"hidden"
],
"valueType": "any",
"label": "",
"value": null
}
],
"outputs": []
},
{
"nodeId": "448745",
"name": "流程开始",
"intro": "",
"avatar": "/imgs/workflow/userChatInput.svg",
"flowNodeType": "workflowStart",
"position": {
"x": 532.1275542407774,
"y": 46.03775600322817
},
"inputs": [
{
"key": "userChatInput",
"renderTypeList": [
"reference",
"textarea"
],
"valueType": "string",
"label": "用户问题",
"required": true,
"toolDescription": "用户问题"
}
],
"outputs": [
{
"id": "userChatInput",
"key": "userChatInput",
"label": "core.module.input.label.user question",
"valueType": "string",
"type": "static"
}
]
},
{
"nodeId": "tMyUnRL5jIrC",
"name": "HTTP 请求",
"intro": "可以发出一个 HTTP 请求，实现更为复杂的操作（联网搜索、数据库查询等）",
"avatar": "/imgs/workflow/http.png",
"flowNodeType": "httpRequest468",
"showStatus": true,
"position": {
"x": 921.2377506442713,
"y": -483.94114977914256
},
"inputs": [
{
"key": "system_addInputParam",
"renderTypeList": [
"addInputParam"
],
"valueType": "dynamic",
"label": "",
"required": false,
"description": "core.module.input.description.HTTP Dynamic Input",
"editField": {
"key": true,
"valueType": true
}
},
{
"key": "prompt",
"valueType": "string",
"label": "prompt",
"renderTypeList": [
"reference"
],
"description": "",
"canEdit": true,
"editField": {
"key": true,
"valueType": true
},
"value": [
"448745",
"userChatInput"
]
},
{
"key": "system_httpMethod",
"renderTypeList": [
"custom"
],
"valueType": "string",
"label": "",
"value": "POST",
"required": true
},
{
"key": "system_httpReqUrl",
"renderTypeList": [
"hidden"
],
"valueType": "string",
"label": "",
"description": "core.module.input.description.Http Request Url",
"placeholder": "https://api.ai.com/getInventory",
"required": false,
"value": "https://api.openai.com/v1/images/generations"
},
{
"key": "system_httpHeader",
"renderTypeList": [
"custom"
],
"valueType": "any",
"value": [
{
"key": "Authorization",
"type": "string",
"value": "Bearer "
}
],
"label": "",
"description": "core.module.input.description.Http Request Header",
"placeholder": "core.module.input.description.Http Request Header",
"required": false
},
{
"key": "system_httpParams",
"renderTypeList": [
"hidden"
],
"valueType": "any",
"value": [],
"label": "",
"required": false
},
{
"key": "system_httpJsonBody",
"renderTypeList": [
"hidden"
],
"valueType": "any",
"value": "{\n \"model\": \"dall-e-3\",\n \"prompt\": \"{{prompt}}\",\n \"n\": 1,\n \"size\": \"1024x1024\"\n}",
"label": "",
"required": false
}
],
"outputs": [
{
"id": "system_addOutputParam",
"key": "system_addOutputParam",
"type": "dynamic",
"valueType": "dynamic",
"label": "",
"editField": {
"key": true,
"valueType": true
}
},
{
"id": "httpRawResponse",
"key": "httpRawResponse",
"label": "原始响应",
"description": "HTTP请求的原始响应。只能接受字符串或JSON类型响应数据。",
"valueType": "any",
"type": "static"
},
{
"id": "DeKGGioBwaMf",
"type": "dynamic",
"key": "data[0].url",
"valueType": "string",
"label": "data[0].url"
}
]
},
{
"nodeId": "CO3POL8svbbi",
"name": "文本加工",
"intro": "可对固定或传入的文本进行加工后输出，非字符串类型数据最终会转成字符串类型。",
"avatar": "/imgs/workflow/textEditor.svg",
"flowNodeType": "pluginModule",
"showStatus": false,
"position": {
"x": 1417.5940290051137,
"y": -478.81889618104356
},
"inputs": [
{
"key": "system_addInputParam",
"valueType": "dynamic",
"label": "动态外部数据",
"renderTypeList": [
"addInputParam"
],
"required": false,
"description": "",
"canEdit": false,
"value": "",
"editField": {
"key": true
},
"dynamicParamDefaultValue": {
"inputType": "reference",
"valueType": "string",
"required": true
}
},
{
"key": "url",
"valueType": "string",
"label": "url",
"renderTypeList": [
"reference"
],
"required": true,
"description": "",
"canEdit": true,
"editField": {
"key": true
},
"value": [
"tMyUnRL5jIrC",
"DeKGGioBwaMf"
]
},
{
"key": "文本",
"valueType": "string",
"label": "文本",
"renderTypeList": [
"textarea"
],
"required": true,
"description": "",
"canEdit": false,
"value": "![]({{url}})",
"editField": {
"key": true
},
"maxLength": "",
"dynamicParamDefaultValue": {
"inputType": "reference",
"valueType": "string",
"required": true
}
}
],
"outputs": [
{
"id": "text",
"type": "static",
"key": "text",
"valueType": "string",
"label": "text",
"description": ""
}
],
"pluginId": "community-textEditor"
},
{
"nodeId": "7mapnCgHfKW6",
"name": "指定回复",
"intro": "该模块可以直接回复一段指定的内容。常用于引导、提示。非字符串内容传入时，会转成字符串进行输出。",
"avatar": "/imgs/workflow/reply.png",
"flowNodeType": "answerNode",
"position": {
"x": 1922.5628399315042,
"y": -471.67391598231796
},
"inputs": [
{
"key": "text",
"renderTypeList": [
"textarea",
"reference"
],
"valueType": "string",
"label": "core.module.input.label.Response content",
"description": "core.module.input.description.Response content",
"placeholder": "core.module.input.description.Response content",
"selectedTypeIndex": 1,
"value": [
"CO3POL8svbbi",
"text"
]
}
],
"outputs": []
}
],
"edges": [
{
"source": "448745",
"target": "tMyUnRL5jIrC",
"sourceHandle": "448745-source-right",
"targetHandle": "tMyUnRL5jIrC-target-left"
},
{
"source": "tMyUnRL5jIrC",
"target": "CO3POL8svbbi",
"sourceHandle": "tMyUnRL5jIrC-source-right",
"targetHandle": "CO3POL8svbbi-target-left"
},
{
"source": "CO3POL8svbbi",
"target": "7mapnCgHfKW6",
"sourceHandle": "CO3POL8svbbi-source-right",
"targetHandle": "7mapnCgHfKW6-target-left"
}
]
}
```
{{% /details %}}

97
docSite/content/zh-cn/docs/use-cases/app-cases/english_essay_correction_bot.md Normal file
View File

@@ -0,0 +1,97 @@
---
title: "英语作文纠错机器人"
description: "使用 FastGPT 创建一个用于英语作文纠错的机器人，帮助用户检测并纠正语言错误"
icon: "spellcheck"
draft: false
toc: true
weight: 608
---
FastGPT 提供了一种基于 LLM Model 搭建应用的简便方式。
本文通过搭建一个英语作文纠错机器人，介绍一下如何使用 **工作流**
## 搭建过程
### 1. 创建工作流
![](/imgs/spellcheck1.png)
可以从 *多轮翻译机器人* 开始创建。
> 多轮翻译机器人是 @米开朗基杨 同学创建的，同样也是一个值得学习的工作流。
### 2. 获取输入，使用大模型进行分析
我们期望让大模型处理文字，返回一个结构化的数据，由我们自己处理。
![](/imgs/spellcheck2.png)
**提示词** 是最重要的一个参数，这里提供的提示词仅供参考：
~~~Markdown
## 角色
资深英语写作专家
## 任务
对输入的原文进行分析。 找出其中的各种错误， 包括但不限于单词拼写错误、 语法错误等。
注意： 忽略标点符号前后空格的问题。
注意： 对于存在错误的句子， 提出修改建议是指指出这个句子中的具体部分， 然后提出将这一个部分修改替换为什么。
## 输出格式
不要使用 Markdown 语法， 输入 JSON 格式的内容。
输出的"reason"的内容使用中文。
直接输出一个列表， 其成员为一个相同类型的对象， 定义如下
您正在找回 FastGPT 账号
```
{
“raw”: string; // 表示原文
“reason”: string; // 表示原因
“suggestion”: string; // 修改建议
}
```
~~~
可以在模型选择的窗口中设置禁用 AI 回复。
这样就看不到输出的 json 格式的内容了。
![](/imgs/spellcheck3.png)
### 3. 数据处理
上面的大模型输出了一个 json，这里要进行数据处理。数据处理可以使用代码执行组件。
![](/imgs/spellcheck4.png)
```JavaScript
function main({data}){
const array = JSON.parse(data)
return {
content: array.map(
(item, index) => {
return `
## 分析${index+1}
- **错误**: ${item.raw}
- **分析**: ${item.reason}
- **修改建议**: ${item.suggestion}
`
}
).join('')
}
}
```
上面的代码将 JSON 解析为 Object, 然后拼接成一串 Markdown 语法的字符串。
FastGPT 的指定回复组件可以将 Markdown 解析为 Html 返回。
## 发布
可以使用发布渠道进行发布。
![](/imgs/spellcheck5.png)
可以选择通过 URL 访问，或者是直接嵌入你的网页中。
> [点我使用](https://share.fastgpt.in/chat/share?shareId=b4r173wkcjae7wpnexcvmyc3)

411
docSite/content/zh-cn/docs/use-cases/app-cases/feishu_webhook.md Normal file
View File

@@ -0,0 +1,411 @@
---
title: '发送飞书webhook通知'
description: '利用工具调用模块，发送一个飞书webhook通知'
icon: 'image'
draft: false
toc: true
weight: 618
---
该文章展示如何发送一个简单的飞书webhook通知，以此类推，发送其他类型的通知也可以这么操作。
| | |
| --------------------- | --------------------- |
| ![](/imgs/feishuwebhook1.webp) | ![](/imgs/feishuwebhook2.webp) |
## 1. 准备飞书机器人
| | | |
| --------------------- | --------------------- |--------------------- |
| ![](/imgs/feishuwebhook3.png) | ![](/imgs/feishuwebhook4.webp) |![](/imgs/feishuwebhook5.png) |
## 2. 导入编排代码
复制下面配置，点击「高级编排」右上角的导入按键，导入该配置，导入后将飞书提供的接口地址复制到「HTTP 模块」。
{{% details title="编排配置" closed="true" %}}
```json
{
"nodes": [
{
"nodeId": "userGuide",
"name": "系统配置",
"intro": "可以配置应用的系统参数",
"avatar": "/imgs/workflow/userGuide.png",
"flowNodeType": "userGuide",
"position": {
"x": 303.41163758039283,
"y": -552.297639861266
},
"version": "481",
"inputs": [],
"outputs": []
},
{
"nodeId": "workflowStartNodeId",
"name": "流程开始",
"intro": "",
"avatar": "/imgs/workflow/userChatInput.svg",
"flowNodeType": "workflowStart",
"position": {
"x": 529.3935295017156,
"y": 197.114018410347
},
"version": "481",
"inputs": [
{
"key": "userChatInput",
"renderTypeList": [
"reference",
"textarea"
],
"valueType": "string",
"label": "用户问题",
"required": true,
"toolDescription": "用户问题"
}
],
"outputs": [
{
"id": "userChatInput",
"key": "userChatInput",
"label": "core.module.input.label.user question",
"valueType": "string",
"type": "static"
}
]
},
{
"nodeId": "u6IAOEssxoZT",
"name": "工具调用",
"intro": "通过AI模型自动选择一个或多个功能块进行调用，也可以对插件进行调用。",
"avatar": "/imgs/workflow/tool.svg",
"flowNodeType": "tools",
"showStatus": true,
"position": {
"x": 1003.146243538873,
"y": 48.52327869406625
},
"version": "481",
"inputs": [
{
"key": "model",
"renderTypeList": [
"settingLLMModel",
"reference"
],
"label": "core.module.input.label.aiModel",
"valueType": "string",
"llmModelType": "all",
"value": "gpt-3.5-turbo"
},
{
"key": "temperature",
"renderTypeList": [
"hidden"
],
"label": "",
"value": 0,
"valueType": "number",
"min": 0,
"max": 10,
"step": 1
},
{
"key": "maxToken",
"renderTypeList": [
"hidden"
],
"label": "",
"value": 2000,
"valueType": "number",
"min": 100,
"max": 4000,
"step": 50
},
{
"key": "systemPrompt",
"renderTypeList": [
"textarea",
"reference"
],
"max": 3000,
"valueType": "string",
"label": "core.ai.Prompt",
"description": "core.app.tip.chatNodeSystemPromptTip",
"placeholder": "core.app.tip.chatNodeSystemPromptTip"
},
{
"key": "history",
"renderTypeList": [
"numberInput",
"reference"
],
"valueType": "chatHistory",
"label": "core.module.input.label.chat history",
"description": "最多携带多少轮对话记录",
"required": true,
"min": 0,
"max": 50,
"value": 6
},
{
"key": "userChatInput",
"renderTypeList": [
"reference",
"textarea"
],
"valueType": "string",
"label": "用户问题",
"required": true,
"value": [
"workflowStartNodeId",
"userChatInput"
]
}
],
"outputs": [
{
"id": "answerText",
"key": "answerText",
"label": "core.module.output.label.Ai response content",
"description": "core.module.output.description.Ai response content",
"valueType": "string",
"type": "static"
}
]
},
{
"nodeId": "fvY5hb0K646V",
"name": "工具调用终止",
"intro": "该模块需配置工具调用使用。当该模块被执行时，本次工具调用将会强制结束，并且不再调用AI针对工具调用结果回答问题。",
"avatar": "/imgs/workflow/toolStop.svg",
"flowNodeType": "stopTool",
"position": {
"x": 2367.838362362707,
"y": 732.355988936165
},
"version": "481",
"inputs": [],
"outputs": []
},
{
"nodeId": "x9rN2a4WnZmt",
"name": "HTTP 请求",
"intro": "向飞书发送一个webhooks通知信息。",
"avatar": "/imgs/workflow/http.png",
"flowNodeType": "httpRequest468",
"showStatus": true,
"position": {
"x": 1623.9214305901633,
"y": 22.777089001645862
},
"version": "486",
"inputs": [
{
"key": "system_addInputParam",
"renderTypeList": [
"addInputParam"
],
"valueType": "dynamic",
"label": "",
"required": false,
"description": "core.module.input.description.HTTP Dynamic Input",
"editField": {
"key": true,
"valueType": true
}
},
{
"valueType": "string",
"renderTypeList": [
"reference"
],
"key": "text",
"label": "text",
"toolDescription": "发送的消息",
"required": true,
"canEdit": true,
"editField": {
"key": true,
"description": true
}
},
{
"key": "system_httpMethod",
"renderTypeList": [
"custom"
],
"valueType": "string",
"label": "",
"value": "POST",
"required": true
},
{
"key": "system_httpReqUrl",
"renderTypeList": [
"hidden"
],
"valueType": "string",
"label": "",
"description": "core.module.input.description.Http Request Url",
"placeholder": "https://api.ai.com/getInventory",
"required": false,
"value": ""
},
{
"key": "system_httpHeader",
"renderTypeList": [
"custom"
],
"valueType": "any",
"value": [],
"label": "",
"description": "core.module.input.description.Http Request Header",
"placeholder": "core.module.input.description.Http Request Header",
"required": false
},
{
"key": "system_httpParams",
"renderTypeList": [
"hidden"
],
"valueType": "any",
"value": [],
"label": "",
"required": false
},
{
"key": "system_httpJsonBody",
"renderTypeList": [
"hidden"
],
"valueType": "any",
"value": "{\r\n \"msg_type\": \"text\",\r\n \"content\": {\r\n \"text\": \"{{text}}\"\r\n }\r\n}",
"label": "",
"required": false
}
],
"outputs": [
{
"id": "system_addOutputParam",
"key": "system_addOutputParam",
"type": "dynamic",
"valueType": "dynamic",
"label": "",
"editField": {
"key": true,
"valueType": true
}
},
{
"id": "error",
"key": "error",
"label": "请求错误",
"description": "HTTP请求错误信息，成功时返回空",
"valueType": "object",
"type": "static"
},
{
"id": "httpRawResponse",
"key": "httpRawResponse",
"label": "原始响应",
"required": true,
"description": "HTTP请求的原始响应。只能接受字符串或JSON类型响应数据。",
"valueType": "any",
"type": "static"
}
]
},
{
"nodeId": "aGHGqH2oUupj",
"name": "指定回复",
"intro": "该模块可以直接回复一段指定的内容。常用于引导、提示。非字符串内容传入时，会转成字符串进行输出。",
"avatar": "/imgs/workflow/reply.png",
"flowNodeType": "answerNode",
"position": {
"x": 2350.7077940158674,
"y": 107.32448732713493
},
"version": "481",
"inputs": [
{
"key": "text",
"renderTypeList": [
"textarea",
"reference"
],
"valueType": "any",
"required": true,
"label": "core.module.input.label.Response content",
"description": "core.module.input.description.Response content",
"placeholder": "core.module.input.description.Response content",
"value": "嘻嘻，发送成功"
}
],
"outputs": []
}
],
"edges": [
{
"source": "workflowStartNodeId",
"target": "u6IAOEssxoZT",
"sourceHandle": "workflowStartNodeId-source-right",
"targetHandle": "u6IAOEssxoZT-target-left"
},
{
"source": "u6IAOEssxoZT",
"target": "x9rN2a4WnZmt",
"sourceHandle": "selectedTools",
"targetHandle": "selectedTools"
},
{
"source": "x9rN2a4WnZmt",
"target": "fvY5hb0K646V",
"sourceHandle": "x9rN2a4WnZmt-source-right",
"targetHandle": "fvY5hb0K646V-target-left"
},
{
"source": "x9rN2a4WnZmt",
"target": "aGHGqH2oUupj",
"sourceHandle": "x9rN2a4WnZmt-source-right",
"targetHandle": "aGHGqH2oUupj-target-left"
}
],
"chatConfig": {
"variables": [
{
"id": "txq1ca",
"key": "test",
"label": "测试",
"type": "custom",
"required": true,
"maxLen": 50,
"enums": [
{
"value": ""
}
]
}
],
"questionGuide": false,
"scheduledTriggerConfig": {
"cronString": "",
"timezone": "Asia/Shanghai",
"defaultPrompt": ""
},
"_id": "66715d4bf577287d39e35ecf"
}
}
```
{{% /details %}}
## 3. 流程说明
1. 为工具调用挂载一个HTTP模块，功能描述写上：调用飞书webhook，发送一个通知。
2. HTTP模块的输入参数中，填写飞书机器人的地址，填写发送的通知内容。
3. HTTP模块输出连接上一个工具终止模块，用于强制结束工具调用。不终止的话，会把调用结果返回给模型，模型会继续回答一次问题，浪费 Tokens
4. HTTP模块输出再连上一个指定回复，直接回复一个发送成功，用于替代AI的回答。

432
docSite/content/zh-cn/docs/use-cases/app-cases/fixingEvidence.md Normal file
View File

@@ -0,0 +1,432 @@
---
title: '固定开头和结尾内容'
description: '利用指定回复，创建固定的开头和结尾'
icon: 'healing'
draft: false
toc: true
weight: 610
---
![](/imgs/demo-fix-evidence1.jpg)
![](/imgs/demo-fix-evidence2.jpg)
如上图，可以通过指定回复编排一个固定的开头和结尾内容。
## 模块编排
复制下面配置，点击「高级编排」右上角的导入按键，导入该配置。
{{% details title="编排配置" closed="true" %}}
```json
{
"nodes": [
{
"nodeId": "7z5g5h",
"name": "流程开始",
"intro": "",
"avatar": "/imgs/workflow/userChatInput.svg",
"flowNodeType": "workflowStart",
"position": {
"x": -269.50851681351924,
"y": 1657.6123698022448
},
"inputs": [
{
"key": "userChatInput",
"renderTypeList": [
"reference",
"textarea"
],
"valueType": "string",
"label": "问题输入",
"required": true,
"toolDescription": "用户问题",
"type": "systemInput",
"showTargetInApp": false,
"showTargetInPlugin": false,
"connected": false,
"selectedTypeIndex": 0,
"value": [
"7z5g5h",
"userChatInput"
]
}
],
"outputs": [
{
"id": "userChatInput",
"type": "static",
"key": "userChatInput",
"valueType": "string",
"label": "core.module.input.label.user question"
}
]
},
{
"nodeId": "nlfwkc",
"name": "AI 对话",
"intro": "AI 大模型对话",
"avatar": "/imgs/workflow/AI.png",
"flowNodeType": "chatNode",
"showStatus": true,
"position": {
"x": 907.2058332478431,
"y": 1348.9992737142143
},
"inputs": [
{
"key": "model",
"renderTypeList": [
"settingLLMModel",
"reference"
],
"label": "core.module.input.label.aiModel",
"valueType": "string",
"type": "selectLLMModel",
"required": true,
"showTargetInApp": false,
"showTargetInPlugin": false,
"value": "gpt-3.5-turbo",
"connected": false,
"selectedTypeIndex": 0
},
{
"key": "temperature",
"renderTypeList": [
"hidden"
],
"label": "",
"value": 0,
"valueType": "number",
"min": 0,
"max": 10,
"step": 1,
"type": "hidden",
"showTargetInApp": false,
"showTargetInPlugin": false,
"connected": false,
"selectedTypeIndex": 0
},
{
"key": "maxToken",
"renderTypeList": [
"hidden"
],
"label": "",
"value": 2000,
"valueType": "number",
"min": 100,
"max": 4000,
"step": 50,
"type": "hidden",
"showTargetInApp": false,
"showTargetInPlugin": false,
"connected": false,
"selectedTypeIndex": 0
},
{
"key": "isResponseAnswerText",
"renderTypeList": [
"hidden"
],
"label": "",
"value": true,
"valueType": "boolean",
"type": "hidden",
"showTargetInApp": false,
"showTargetInPlugin": false,
"connected": false,
"selectedTypeIndex": 0
},
{
"key": "quoteTemplate",
"renderTypeList": [
"hidden"
],
"label": "",
"valueType": "string",
"type": "hidden",
"showTargetInApp": false,
"showTargetInPlugin": false,
"connected": false,
"selectedTypeIndex": 0
},
{
"key": "quotePrompt",
"renderTypeList": [
"hidden"
],
"label": "",
"valueType": "string",
"type": "hidden",
"showTargetInApp": false,
"showTargetInPlugin": false,
"connected": false,
"selectedTypeIndex": 0
},
{
"key": "systemPrompt",
"renderTypeList": [
"textarea",
"reference"
],
"max": 300,
"valueType": "string",
"label": "core.ai.Prompt",
"description": "core.app.tip.chatNodeSystemPromptTip",
"placeholder": "core.app.tip.chatNodeSystemPromptTip",
"type": "textarea",
"showTargetInApp": true,
"showTargetInPlugin": true,
"value": "",
"connected": false,
"selectedTypeIndex": 0
},
{
"key": "history",
"renderTypeList": [
"numberInput",
"reference"
],
"valueType": "chatHistory",
"label": "core.module.input.label.chat history",
"required": true,
"min": 0,
"max": 30,
"value": 6,
"type": "numberInput",
"showTargetInApp": true,
"showTargetInPlugin": true,
"connected": false,
"selectedTypeIndex": 0
},
{
"key": "userChatInput",
"renderTypeList": [
"reference",
"textarea"
],
"valueType": "string",
"label": "问题输入",
"required": true,
"toolDescription": "用户问题",
"type": "custom",
"showTargetInApp": true,
"showTargetInPlugin": true,
"connected": true,
"selectedTypeIndex": 0,
"value": [
"7z5g5h",
"userChatInput"
]
},
{
"key": "quoteQA",
"renderTypeList": [
"settingDatasetQuotePrompt"
],
"label": "",
"debugLabel": "知识库引用",
"description": "core.module.Dataset quote.Input description",
"valueType": "datasetQuote",
"type": "target",
"showTargetInApp": true,
"showTargetInPlugin": true,
"connected": true,
"selectedTypeIndex": 0,
"value": [
"fljhzy",
"quoteQA"
]
}
],
"outputs": [
{
"id": "answerText",
"type": "static",
"key": "answerText",
"valueType": "string",
"label": "core.module.output.label.Ai response content",
"description": "core.module.output.description.Ai response content"
},
{
"id": "history",
"type": "static",
"key": "history",
"valueType": "chatHistory",
"label": "core.module.output.label.New context",
"description": "core.module.output.description.New context"
}
]
},
{
"nodeId": "q9equb",
"name": "core.module.template.App system setting",
"intro": "可以配置应用的系统参数。",
"avatar": "/imgs/workflow/userGuide.png",
"flowNodeType": "userGuide",
"position": {
"x": -275.92529567956024,
"y": 1094.1001488133452
},
"inputs": [
{
"key": "welcomeText",
"renderTypeList": [
"hidden"
],
"valueType": "string",
"label": "core.app.Welcome Text",
"type": "hidden",
"showTargetInApp": false,
"showTargetInPlugin": false,
"value": "你好，我是电影《星际穿越》 AI 助手，有什么可以帮助你的？\n[导演是谁]\n[剧情介绍]\n[票房分析]",
"connected": false,
"selectedTypeIndex": 0
},
{
"key": "variables",
"renderTypeList": [
"hidden"
],
"valueType": "any",
"label": "core.module.Variable",
"value": [],
"type": "hidden",
"showTargetInApp": false,
"showTargetInPlugin": false,
"connected": false,
"selectedTypeIndex": 0
},
{
"key": "questionGuide",
"valueType": "boolean",
"renderTypeList": [
"hidden"
],
"label": "",
"type": "switch",
"showTargetInApp": false,
"showTargetInPlugin": false,
"connected": false,
"selectedTypeIndex": 0
},
{
"key": "tts",
"renderTypeList": [
"hidden"
],
"valueType": "any",
"label": "",
"type": "hidden",
"showTargetInApp": false,
"showTargetInPlugin": false,
"connected": false,
"selectedTypeIndex": 0
},
{
"key": "whisper",
"renderTypeList": [
"hidden"
],
"valueType": "any",
"label": ""
},
{
"key": "scheduleTrigger",
"renderTypeList": [
"hidden"
],
"valueType": "any",
"label": "",
"value": null
}
],
"outputs": []
},
{
"nodeId": "tc90wz",
"name": "指定回复",
"intro": "该模块可以直接回复一段指定的内容。常用于引导、提示。非字符串内容传入时，会转成字符串进行输出。",
"avatar": "/imgs/workflow/reply.png",
"flowNodeType": "answerNode",
"position": {
"x": 159.49274056478237,
"y": 1621.4635230667668
},
"inputs": [
{
"key": "text",
"renderTypeList": [
"textarea",
"reference"
],
"valueType": "any",
"label": "core.module.input.label.Response content",
"description": "core.module.input.description.Response content",
"placeholder": "core.module.input.description.Response content",
"type": "textarea",
"showTargetInApp": true,
"showTargetInPlugin": true,
"value": "这是开头\\n",
"connected": false,
"selectedTypeIndex": 0
}
],
"outputs": []
},
{
"nodeId": "U5T3dMVY4wj7",
"name": "指定回复",
"intro": "该模块可以直接回复一段指定的内容。常用于引导、提示。非字符串内容传入时，会转成字符串进行输出。",
"avatar": "/imgs/workflow/reply.png",
"flowNodeType": "answerNode",
"position": {
"x": 1467.0625486167608,
"y": 1597.346243737531
},
"inputs": [
{
"key": "text",
"renderTypeList": [
"textarea",
"reference"
],
"valueType": "string",
"label": "core.module.input.label.Response content",
"description": "core.module.input.description.Response content",
"placeholder": "core.module.input.description.Response content",
"value": "这是结尾"
}
],
"outputs": []
}
],
"edges": [
{
"source": "7z5g5h",
"target": "tc90wz",
"sourceHandle": "7z5g5h-source-right",
"targetHandle": "tc90wz-target-left"
},
{
"source": "tc90wz",
"target": "nlfwkc",
"sourceHandle": "tc90wz-source-right",
"targetHandle": "nlfwkc-target-left"
},
{
"source": "nlfwkc",
"target": "U5T3dMVY4wj7",
"sourceHandle": "nlfwkc-source-right",
"targetHandle": "U5T3dMVY4wj7-target-left"
}
]
}
```
{{% /details %}}

1285
docSite/content/zh-cn/docs/use-cases/app-cases/google_search.md Normal file
View File

File diff suppressed because it is too large Load Diff

853
docSite/content/zh-cn/docs/use-cases/app-cases/lab_appointment.md Normal file
View File

@@ -0,0 +1,853 @@
---
title: '实验室预约'
description: '展示高级编排操作数据库的能力'
icon: 'database'
draft: false
toc: true
weight: 612
---
| | |
| --------------------- | --------------------- |
| ![](/imgs/demo-appointment1.webp) | ![](/imgs/demo-appointment2.webp) |
| ![](/imgs/demo-appointment3.webp) | ![](/imgs/demo-appointment4.webp) |
本示例演示了利用工具调用，自动选择调用知识库搜索实验室相关内容，或调用 HTTP 模块实现数据库的 CRUD 操作。
以一个实验室预约为例，用户可以通过对话系统预约、取消、修改预约和查询预约记录。
## 1. 全局变量使用
通过设计一个全局变量，让用户输入姓名，模拟用户身份信息。实际使用过程中，通常是直接通过嵌入 Token 来标记用户身份。
## 2. 工具调用
![](/imgs/demo-appointment5.png)
背景知识中，引导模型调用工具去执行不通的操作。
{{% alert icon="🤗" context="warning" %}}
**Tips:** 这里需要增加适当的上下文，方便模型结合历史纪录进行判断和决策~
{{% /alert %}}
## 3. HTTP 模块
![](/imgs/demo-appointment6.jpg)
HTTP模块中，需要设置 3 个工具参数：
- 预约行为：可取 get, put, post, delete 四个值，分别对应查询、修改、新增、删除操作。当然，你也可以写4个HTTP模块，来分别处理。
- labname: 实验室名。非必填，因为查询和删除时候，不需要。
- time: 预约时间。
# 总结
1. 工具调用模块是非常强大的功能，可以在一定程度上替代问题分类和内容提取。
2. 通过工具模块，动态的调用不同的工具，可以将复杂业务解耦。
# 附件
## 编排配置
可直接复制，导入到 FastGPT 中。
{{% details title="编排配置" closed="true" %}}
```json
{
"nodes": [
{
"nodeId": "userChatInput",
"name": "流程开始",
"intro": "当用户发送一个内容后，流程将会从这个模块开始执行。",
"avatar": "/imgs/workflow/userChatInput.svg",
"flowNodeType": "workflowStart",
"position": {
"x": 309.7143912167367,
"y": 1501.2761754220846
},
"inputs": [
{
"key": "userChatInput",
"renderTypeList": [
"reference",
"textarea"
],
"valueType": "string",
"label": "问题输入",
"required": true,
"toolDescription": "用户问题",
"type": "systemInput",
"showTargetInApp": false,
"showTargetInPlugin": false,
"connected": false,
"selectedTypeIndex": 0,
"value": [
"userChatInput",
"userChatInput"
]
}
],
"outputs": [
{
"id": "userChatInput",
"type": "static",
"key": "userChatInput",
"valueType": "string",
"label": "core.module.input.label.user question"
}
]
},
{
"nodeId": "eg5upi",
"name": "指定回复",
"intro": "该模块可以直接回复一段指定的内容。常用于引导、提示。非字符串内容传入时，会转成字符串进行输出。",
"avatar": "/imgs/workflow/reply.png",
"flowNodeType": "answerNode",
"position": {
"x": 1962.729630445213,
"y": 2295.9791334948304
},
"inputs": [
{
"key": "text",
"renderTypeList": [
"textarea",
"reference"
],
"valueType": "any",
"label": "core.module.input.label.Response content",
"description": "core.module.input.description.Response content",
"placeholder": "core.module.input.description.Response content",
"type": "textarea",
"showTargetInApp": true,
"showTargetInPlugin": true,
"connected": true,
"selectedTypeIndex": 1,
"value": [
"40clf3",
"result"
]
}
],
"outputs": []
},
{
"nodeId": "kge59i",
"name": "用户引导",
"intro": "可以配置应用的系统参数。",
"avatar": "/imgs/workflow/userGuide.png",
"flowNodeType": "userGuide",
"position": {
"x": -327.218389965887,
"y": 1504.8056414948464
},
"inputs": [
{
"key": "welcomeText",
"renderTypeList": [
"hidden"
],
"valueType": "string",
"label": "core.app.Welcome Text",
"type": "hidden",
"showTargetInApp": false,
"showTargetInPlugin": false,
"value": "你好，我是实验室助手，请问有什么可以帮助你的么？如需预约或修改预约实验室，请提供姓名、时间和实验室名称。\n[实验室介绍]\n[开放时间]\n[预约]",
"connected": false,
"selectedTypeIndex": 0
},
{
"key": "variables",
"renderTypeList": [
"hidden"
],
"valueType": "any",
"label": "core.module.Variable",
"value": [
{
"id": "gt9b23",
"key": "name",
"label": "name",
"type": "input",
"required": true,
"maxLen": 50,
"enums": [
{
"value": ""
}
]
}
],
"type": "hidden",
"showTargetInApp": false,
"showTargetInPlugin": false,
"connected": false,
"selectedTypeIndex": 0
},
{
"key": "questionGuide",
"valueType": "boolean",
"renderTypeList": [
"hidden"
],
"label": "",
"type": "switch",
"showTargetInApp": false,
"showTargetInPlugin": false,
"value": false,
"connected": false,
"selectedTypeIndex": 0
},
{
"key": "tts",
"renderTypeList": [
"hidden"
],
"valueType": "any",
"label": "",
"type": "hidden",
"showTargetInApp": false,
"showTargetInPlugin": false,
"value": {
"type": "model",
"model": "tts-1",
"voice": "alloy"
},
"connected": false,
"selectedTypeIndex": 0
},
{
"key": "whisper",
"renderTypeList": [
"hidden"
],
"valueType": "any",
"label": "",
"type": "hidden",
"showTargetInApp": false,
"showTargetInPlugin": false,
"connected": false,
"selectedTypeIndex": 0
},
{
"key": "scheduleTrigger",
"renderTypeList": [
"hidden"
],
"valueType": "any",
"label": "",
"value": null
}
],
"outputs": []
},
{
"nodeId": "40clf3",
"name": "HTTP请求",
"intro": "可以发出一个 HTTP 请求，实现更为复杂的操作（联网搜索、数据库查询等）",
"avatar": "/imgs/workflow/http.png",
"flowNodeType": "httpRequest468",
"showStatus": true,
"position": {
"x": 1118.6532653446993,
"y": 1955.886106913907
},
"inputs": [
{
"key": "system_httpMethod",
"renderTypeList": [
"custom"
],
"valueType": "string",
"label": "",
"value": "POST",
"required": true,
"type": "custom",
"showTargetInApp": false,
"showTargetInPlugin": false,
"connected": false,
"selectedTypeIndex": 0
},
{
"valueType": "string",
"renderTypeList": [
"reference"
],
"key": "action",
"label": "action",
"toolDescription": "预约行为，一共四种：\nget - 查询预约情况\nput - 更新预约\npost - 新增预约\ndelete - 删除预约",
"required": true,
"canEdit": true,
"editField": {
"key": true,
"description": true
}
},
{
"valueType": "string",
"renderTypeList": [
"reference"
],
"key": "labname",
"label": "labname",
"toolDescription": "实验室名称",
"required": false,
"canEdit": true,
"editField": {
"key": true,
"description": true
}
},
{
"valueType": "string",
"renderTypeList": [
"reference"
],
"key": "time",
"label": "time",
"toolDescription": "预约时间，按 YYYY/MM/DD HH:mm 格式返回",
"required": false,
"canEdit": true,
"editField": {
"key": true,
"description": true
}
},
{
"key": "system_httpReqUrl",
"renderTypeList": [
"hidden"
],
"valueType": "string",
"label": "",
"description": "core.module.input.description.Http Request Url",
"placeholder": "https://api.ai.com/getInventory",
"required": false,
"type": "hidden",
"showTargetInApp": false,
"showTargetInPlugin": false,
"value": "https://d8dns0.laf.dev/appointment-lab",
"connected": false,
"selectedTypeIndex": 0
},
{
"key": "system_httpHeader",
"renderTypeList": [
"custom"
],
"valueType": "any",
"value": [],
"label": "",
"description": "core.module.input.description.Http Request Header",
"placeholder": "core.module.input.description.Http Request Header",
"required": false,
"type": "custom",
"showTargetInApp": false,
"showTargetInPlugin": false,
"connected": false,
"selectedTypeIndex": 0
},
{
"key": "system_httpParams",
"renderTypeList": [
"hidden"
],
"valueType": "any",
"value": [],
"label": "",
"required": false,
"type": "hidden",
"showTargetInApp": false,
"showTargetInPlugin": false,
"connected": false,
"selectedTypeIndex": 0
},
{
"key": "system_httpJsonBody",
"renderTypeList": [
"hidden"
],
"valueType": "any",
"value": "{\r\n \"name\": \"{{name}}\",\r\n \"time\": \"{{time}}\",\r\n \"labname\": \"{{labname}}\",\r\n \"action\": \"{{action}}\"\r\n}",
"label": "",
"required": false,
"type": "hidden",
"showTargetInApp": false,
"showTargetInPlugin": false,
"connected": false,
"selectedTypeIndex": 0
},
{
"key": "system_addInputParam",
"renderTypeList": [
"addInputParam"
],
"valueType": "dynamic",
"label": "",
"required": false,
"description": "core.module.input.description.HTTP Dynamic Input",
"editField": {
"key": true,
"valueType": true
}
}
],
"outputs": [
{
"id": "system_addOutputParam",
"type": "dynamic",
"key": "system_addOutputParam",
"valueType": "dynamic",
"label": "",
"editField": {
"key": true,
"valueType": true
}
},
{
"id": "result",
"type": "static",
"key": "result",
"valueType": "string",
"label": "result",
"description": "result",
"canEdit": true,
"editField": {
"key": true,
"name": true,
"description": true,
"dataType": true
}
},
{
"id": "httpRawResponse",
"type": "static",
"key": "httpRawResponse",
"valueType": "any",
"label": "原始响应",
"description": "HTTP请求的原始响应。只能接受字符串或JSON类型响应数据。"
}
]
},
{
"nodeId": "fYxwWym8flYL",
"name": "工具调用",
"intro": "通过AI模型自动选择一个或多个功能块进行调用，也可以对插件进行调用。",
"avatar": "/imgs/workflow/tool.svg",
"flowNodeType": "tools",
"showStatus": true,
"position": {
"x": 933.9342354248961,
"y": 1229.3563445150553
},
"inputs": [
{
"key": "model",
"renderTypeList": [
"settingLLMModel",
"reference"
],
"label": "core.module.input.label.aiModel",
"valueType": "string",
"llmModelType": "all",
"value": "gpt-3.5-turbo"
},
{
"key": "temperature",
"renderTypeList": [
"hidden"
],
"label": "",
"value": 0,
"valueType": "number",
"min": 0,
"max": 10,
"step": 1
},
{
"key": "maxToken",
"renderTypeList": [
"hidden"
],
"label": "",
"value": 2000,
"valueType": "number",
"min": 100,
"max": 4000,
"step": 50
},
{
"key": "systemPrompt",
"renderTypeList": [
"textarea",
"reference"
],
"max": 3000,
"valueType": "string",
"label": "core.ai.Prompt",
"description": "core.app.tip.chatNodeSystemPromptTip",
"placeholder": "core.app.tip.chatNodeSystemPromptTip",
"value": "当前时间为: {{cTime}}\n你是实验室助手，用户可能会询问实验室相关介绍或预约实验室。\n请选择合适的工具去帮助他们。"
},
{
"key": "history",
"renderTypeList": [
"numberInput",
"reference"
],
"valueType": "chatHistory",
"label": "core.module.input.label.chat history",
"required": true,
"min": 0,
"max": 30,
"value": 6
},
{
"key": "userChatInput",
"renderTypeList": [
"reference",
"textarea"
],
"valueType": "string",
"label": "用户问题",
"required": true,
"value": [
"userChatInput",
"userChatInput"
]
}
],
"outputs": []
},
{
"nodeId": "JSSQtDgwmmbE",
"name": "知识库搜索",
"intro": "调用“语义检索”和“全文检索”能力，从“知识库”中查找实验室介绍和使用规则等信息。",
"avatar": "/imgs/workflow/db.png",
"flowNodeType": "datasetSearchNode",
"showStatus": true,
"position": {
"x": 447.0795498711184,
"y": 1971.5311041711186
},
"inputs": [
{
"key": "datasets",
"renderTypeList": [
"selectDataset",
"reference"
],
"label": "core.module.input.label.Select dataset",
"value": [],
"valueType": "selectDataset",
"list": [],
"required": true
},
{
"key": "similarity",
"renderTypeList": [
"selectDatasetParamsModal"
],
"label": "",
"value": 0.4,
"valueType": "number"
},
{
"key": "limit",
"renderTypeList": [
"hidden"
],
"label": "",
"value": 1500,
"valueType": "number"
},
{
"key": "searchMode",
"renderTypeList": [
"hidden"
],
"label": "",
"valueType": "string",
"value": "embedding"
},
{
"key": "usingReRank",
"renderTypeList": [
"hidden"
],
"label": "",
"valueType": "boolean",
"value": false
},
{
"key": "datasetSearchUsingExtensionQuery",
"renderTypeList": [
"hidden"
],
"label": "",
"valueType": "boolean",
"value": false
},
{
"key": "datasetSearchExtensionModel",
"renderTypeList": [
"hidden"
],
"label": "",
"valueType": "string",
"value": "gpt-3.5-turbo"
},
{
"key": "datasetSearchExtensionBg",
"renderTypeList": [
"hidden"
],
"label": "",
"valueType": "string",
"value": ""
},
{
"key": "userChatInput",
"renderTypeList": [
"reference",
"textarea"
],
"valueType": "string",
"label": "用户问题",
"required": true,
"toolDescription": "需要检索的内容"
}
],
"outputs": [
{
"id": "quoteQA",
"key": "quoteQA",
"label": "core.module.Dataset quote.label",
"description": "特殊数组格式，搜索结果为空时，返回空数组。",
"type": "static",
"valueType": "datasetQuote"
}
]
},
{
"nodeId": "IdntVQiTopHT",
"name": "工具调用终止",
"intro": "该模块需配置工具调用使用。当该模块被执行时，本次工具调用将会强制结束，并且不再调用AI针对工具调用结果回答问题。",
"avatar": "/imgs/workflow/toolStop.svg",
"flowNodeType": "stopTool",
"position": {
"x": 1969.73331750207,
"y": 2650.0258908119413
},
"inputs": [],
"outputs": []
}
],
"edges": [
{
"source": "40clf3",
"target": "eg5upi",
"sourceHandle": "40clf3-source-right",
"targetHandle": "eg5upi-target-left"
},
{
"source": "userChatInput",
"target": "fYxwWym8flYL",
"sourceHandle": "userChatInput-source-right",
"targetHandle": "fYxwWym8flYL-target-left"
},
{
"source": "fYxwWym8flYL",
"target": "40clf3",
"sourceHandle": "selectedTools",
"targetHandle": "selectedTools"
},
{
"source": "fYxwWym8flYL",
"target": "JSSQtDgwmmbE",
"sourceHandle": "selectedTools",
"targetHandle": "selectedTools"
},
{
"source": "40clf3",
"target": "IdntVQiTopHT",
"sourceHandle": "40clf3-source-right",
"targetHandle": "IdntVQiTopHT-target-left"
}
]
}
```
{{% /details %}}
## Laf 云函数代码
可以在 [Laf](https://laf.dev/) 中快速构建 HTTP 接口。
{{% details title="函数代码" closed="true" %}}
```ts
import cloud from '@lafjs/cloud'
const db = cloud.database()
type RequestType = {
name: string;
time?: string;
labname?: string;
action: 'post' | 'delete' | 'put' | 'get'
}
export default async function (ctx: FunctionContext) {
try {
const { action,...body } = ctx.body as RequestType
if (action === 'get') {
return await getRecord(ctx.body)
}
if (action === 'post') {
return await createRecord(ctx.body)
}
if (action === 'put') {
return await putRecord(ctx.body)
}
if (action === 'delete') {
return await removeRecord(ctx.body)
}
return {
result: "异常"
}
} catch (err) {
return {
result: "异常"
}
}
}
async function putRecord({ name, time, labname }: RequestType) {
const missData = []
if (!name) missData.push("你的姓名")
if (missData.length > 0) {
return {
result: `请提供: ${missData.join("、")}`
}
}
const { data: record } = await db.collection("LabAppointment").where({
name, status: "unStart"
}).getOne()
if (!record) {
return {
result: `${name} 还没有预约记录`
}
}
const updateWhere = {
name,
time: time || record.time,
labname: labname || record.labname
}
await db.collection("LabAppointment").where({
name, status: "unStart"
}).update(updateWhere)
return {
result: `修改预约成功。
姓名：${name}·
时间: ${updateWhere.time}
实验室名: ${updateWhere.labname}
` }
}
async function getRecord({ name }: RequestType) {
if (!name) {
return {
result: "请提供你的姓名"
}
}
const { data } = await db.collection('LabAppointment').where({ name, status: "unStart" }).getOne()
if (!data) {
return {
result: `${name} 没有预约中的记录`
}
}
return {
result: `${name} 有一条预约记录：
姓名：${data.name}
时间: ${data.time}
实验室名: ${data.labname}
`
}
}
async function removeRecord({ name }: RequestType) {
if (!name) {
return {
result: "请提供你的姓名"
}
}
const { deleted } = await db.collection('LabAppointment').where({ name, status: "unStart" }).remove()
if (deleted > 0) {
return {
result: `取消预约记录成功: ${name}`
}
}
return {
result: ` ${name} 没有预约中的记录`
}
}
async function createRecord({ name, time, labname }: RequestType) {
const missData = []
if (!name) missData.push("你的姓名")
if (!time) missData.push("需要预约的时间")
if (!labname) missData.push("实验室名名称")
if (missData.length > 0) {
return {
result: `请提供: ${missData.join("、")}`
}
}
const { data: record } = await db.collection("LabAppointment").where({
name, status: "unStart"
}).getOne()
if (record) {
return {
result: `您已经有一个预约记录了:
姓名：${record.name}
时间: ${record.time}
实验室名: ${record.labname}
每人仅能同时预约一个实验室名。
`
}
}
await db.collection("LabAppointment").add({
name, time, labname, status: "unStart"
})
return {
result: `预约成功。
姓名：${name}
时间: ${time}
实验室名: ${labname}
` }
}
```
{{% /details %}}

315
docSite/content/zh-cn/docs/use-cases/app-cases/multi_turn_translation_bot.md Normal file
View File

@@ -0,0 +1,315 @@
---
title: "多轮翻译机器人"
description: "如何使用 FastGPT 构建一个多轮翻译机器人，实现连续的对话翻译功能"
icon: "translate"
draft: false
toc: true
weight: 606
---
吴恩达老师提出了一种反思翻译的大语言模型(LLM)翻译工作流程——[GitHub - andrewyng/translation-agent](https://github.com/andrewyng/translation-agent)，具体工作流程如下：
1. 提示一个 LLM 将文本从 `source_language` 翻译到 `target_language`；
2. 让 LLM 反思翻译结果并提出建设性的改进建议；
3. 使用这些建议来改进翻译。
这个翻译流程应该是目前比较新的一种翻译方式，利用 LLM 对自己的翻译结果进行改进来获得较好的翻译效果
项目中展示了可以利用对长文本进行分片，然后分别进行反思翻译处理，以突破 LLM 对 tokens 数量的限制，真正实现长文本一键高效率高质量翻译。
项目还通过给大模型限定国家地区，已实现更精确的翻译，如美式英语、英式英语之分；同时提出一些可能能带来更好效果的优化，如对于一些 LLM 未曾训练到的术语（或有多种翻译方式的术语）建立术语表，进一步提升翻译的精确度等等
而这一切都能通过 Fastgpt 工作流轻松实现，本文将手把手教你如何复刻吴恩达老师的 translation-agent
# 单文本块反思翻译
先从简单的开始，即不超出 LLM tokens 数量限制的单文本块翻译
## 初始翻译
第一步先让 LLM 对源文本块进行初始翻译（翻译的提示词在源项目中都有）
![](/imgs/translate1.png)
通过`文本拼接`模块引用 源语言、目标语言、源文本这三个参数，生成提示词，传给 LLM，让它给出第一版的翻译
## 反思
然后让 LLM 对第一步生成的初始翻译给出修改建议，称之为 反思
![](/imgs/translate2.png)
这时的提示词接收 5 个参数，源文本、初始翻译、源语言、目标语言 以及限定词地区国家，这样 LLM 会对前面生成的翻译提出相当多的修改建议，为后续的提升翻译作准备
## 提升翻译
![](/imgs/translate3.png)
在前文生成了初始翻译以及相应的反思后，将这二者输入给第三次 LLM 翻译，这样我们就能获得一个比较高质量的翻译结果
完整的工作流如下
![](/imgs/translate4.png)
## 运行效果
由于考虑之后对这个反思翻译的复用，所以创建了一个插件，那么在下面我直接调用这个插件就能使用反思翻译，效果如下
随机挑选了一段哈利波特的文段
![](/imgs/translate5.png)
![](/imgs/translate6.png)
可以看到反思翻译后的效果还是好上不少的，其中反思的输出如下
![](/imgs/translate7.png)
# 长文反思翻译
在掌握了对短文本块的反思翻译后，我们能轻松的通过分片和循环，实现对长文本也即多文本块的反思翻译
整体的逻辑是，首先对传入文本的 tokens数量做判断，如果不超过设置的 tokens 限制，那么直接调用单文本块反思翻译，如果超过设置的 tokens限制，那么切割为合理的大小，再分别进行对应的反思翻译处理
## 计算 tokens
![](/imgs/translate8.png)
首先，我使用了 Laf函数 模块来实现对输入文本的 tokens 的计算
laf函数的使用相当简单，即开即用，只需要在 laf 创建个应用，然后安装 tiktoken 依赖，导入如下代码即可
```TypeScript
const { Tiktoken } = require("tiktoken/lite");
const cl100k_base = require("tiktoken/encoders/cl100k_base.json");
interface IRequestBody {
str: string
}
interface RequestProps extends IRequestBody {
systemParams: {
appId: string,
variables: string,
histories: string,
cTime: string,
chatId: string,
responseChatItemId: string
}
}
interface IResponse {
message: string;
tokens: number;
}
export default async function (ctx: FunctionContext): Promise<IResponse> {
const { str = "" }: RequestProps = ctx.body
const encoding = new Tiktoken(
cl100k_base.bpe_ranks,
cl100k_base.special_tokens,
cl100k_base.pat_str
);
const tokens = encoding.encode(str);
encoding.free();
return {
message: 'ok',
tokens: tokens.length
};
}
```
再回到 Fastgpt，点击“同步参数”，再连线将源文本传入，即可计算 tokens 数量
## 计算单文本块大小
![](/imgs/translate9.png)
由于不涉及第三方包，只是一些数据处理，所以直接使用 代码运行 模块处理即可
```TypeScript
function main({tokenCount, tokenLimit}){
const numChunks = Math.ceil(tokenCount / tokenLimit);
let chunkSize = Math.floor(tokenCount / numChunks);
const remainingTokens = tokenCount % tokenLimit;
if (remainingTokens > 0) {
chunkSize += Math.floor(remainingTokens / numChunks);
}
return {chunkSize};
}
```
通过上面的代码，我们就能算出不超过 token限制的合理单文本块大小是多少了
## 获得切分后源文本块
![](/imgs/translate10.png)
通过单文本块大小和源文本，我们再编写一个函数调用 langchain 的 textsplitters 包来实现文本分片，具体代码如下
```TypeScript
import cloud from '@lafjs/cloud'
import { TokenTextSplitter } from "@langchain/textsplitters";
interface IRequestBody {
text: string
chunkSize: number
}
interface RequestProps extends IRequestBody {
systemParams: {
appId: string,
variables: string,
histories: string,
cTime: string,
chatId: string,
responseChatItemId: string
}
}
interface IResponse {
output: string[];
}
export default async function (ctx: FunctionContext): Promise<IResponse>{
const { text = '', chunkSize=1000 }: RequestProps = ctx.body;
const splitter = new TokenTextSplitter({
encodingName:"gpt2",
chunkSize: Number(chunkSize),
chunkOverlap: 0,
});
const output = await splitter.splitText(text);
return {
output
}
}
```
这样我们就获得了切分好的文本，接下去的操作就类似单文本块反思翻译
## 多文本块翻译
这里应该还是不能直接调用前面的单文本块反思翻译，因为提示词中会涉及一些上下文的处理（或者可以修改下前面写好的插件，多传点参数进去）
详细的和前面类似，就是提示词进行一些替换，以及需要做一些很简单的数据处理，整体效果如下
### 多文本块初始翻译
![](/imgs/translate11.png)
### 多文本块反思
![](/imgs/translate12.png)
### 多文本块提升翻译
![](/imgs/translate13.png)
## 循环执行
长文反思翻译比较关键的一个部分，就是对多个文本块进行循环反思翻译
Fastgpt 提供了工作流线路可以返回去执行的功能，所以我们可以写一个很简单的判断函数，来判断结束或是接着执行
![](/imgs/translate14.png)
也就是通过判断当前处理的这个文本块，是否是最后一个文本块，从而判断是否需要继续执行，就这样，我们实现了长文反思翻译的效果
完整工作流如下
![](/imgs/translate15.png)
## 运行效果
首先输入全局设置
![](/imgs/translate16.png)
然后输入需要翻译的文本，这里我选择了一章哈利波特的英文原文来做翻译，其文本长度通过 openai 对 tokens 数量的判断如下
![](/imgs/translate17.png)
实际运行效果如下
![](/imgs/translate18.png)
可以看到还是能满足阅读需求的
# 进一步调优
## 提示词调优
在源项目中，给 AI 的系统提示词还是比较的简略的，我们可以通过比较完善的提示词，来督促 LLM 返回更合适的翻译，进一步提升翻译的质量
比如初始翻译中，
```TypeScript
# Role: 资深翻译专家
## Background:
你是一位经验丰富的翻译专家,精通{{source_lang}}和{{target_lang}}互译,尤其擅长将{{source_lang}}文章译成流畅易懂的{{target_lang}}。你曾多次带领团队完成大型翻译项目,译文广受好评。
## Attention:
- 翻译过程中要始终坚持"信、达、雅"的原则,但"达"尤为重要
- 译文要符合{{target_lang}}的表达习惯,通俗易懂,连贯流畅
- 避免使用过于文绉绉的表达和晦涩难懂的典故引用
## Constraints:
- 必须严格遵循四轮翻译流程:直译、意译、校审、定稿
- 译文要忠实原文,准确无误,不能遗漏或曲解原意
## Goals:
- 通过四轮翻译流程,将{{source_lang}}原文译成高质量的{{target_lang}}译文
- 译文要准确传达原文意思,语言表达力求浅显易懂,朗朗上口
- 适度使用一些熟语俗语、流行网络用语等,增强译文的亲和力
- 在直译的基础上,提供至少2个不同风格的意译版本供选择
## Skills:
- 精通{{source_lang}} {{target_lang}}两种语言,具有扎实的语言功底和丰富的翻译经验
- 擅长将{{source_lang}}表达习惯转换为地道自然的{{target_lang}}
- 对当代{{target_lang}}语言的发展变化有敏锐洞察,善于把握语言流行趋势
## Workflow:
1. 第一轮直译:逐字逐句忠实原文,不遗漏任何信息
2. 第二轮意译:在直译的基础上用通俗流畅的{{target_lang}}意译原文,至少提供2个不同风格的版本
3. 第三轮校审:仔细审视译文,消除偏差和欠缺,使译文更加地道易懂
4. 第四轮定稿:择优选取,反复修改润色,最终定稿出一个简洁畅达、符合大众阅读习惯的译文
## OutputFormat:
- 只需要输出第四轮定稿的回答
## Suggestions:
- 直译时力求忠实原文,但不要过于拘泥逐字逐句
- 意译时在准确表达原意的基础上,用最朴实无华的{{target_lang}}来表达
- 校审环节重点关注译文是否符合{{target_lang}}表达习惯,是否通俗易懂
- 定稿时适度采用一些熟语谚语、网络流行语等,使译文更接地气- 善于利用{{target_lang}}的灵活性,用不同的表述方式展现同一内容,提高译文的可读性
```
从而返回更准确更高质量的初始翻译，后续的反思和提升翻译也可以修改更准确的提示词，如下
![](/imgs/translate19.png)
然后再让我们来看看运行效果
![](/imgs/translate20.png)
给了和之前相同的一段文本进行测试，测试效果还是比较显著的，就比如红框部分，之前的翻译如下
![](/imgs/translate21.png)
从“让你的猫头鹰给我写信”这样有失偏颇的翻译，变成“给我写信，你的猫头鹰会知道怎么找到我”这样较为准确的翻译
## 其他调优
比如限定词调优，源项目中已经做了示范，就是加上国家地区这个限定词，实测确实会有不少提升
出于 LLM 的卓越能力，我们能够通过设置不同的prompt来获取不同的翻译结果，也就是可以很轻松地通过设置特殊的限定词，来实现特定的，更精确的翻译
而对于一些超出 LLM 理解的术语等，也可以利用 Fastgpt 的知识库功能进行相应扩展，进一步完善翻译机器人的功能

86
docSite/content/zh-cn/docs/use-cases/app-cases/submit_application_template.md Normal file
View File

@@ -0,0 +1,86 @@
---
title: "如何提交应用模板"
description: "指南：如何向 FastGPT 提交应用模板"
icon: "template_submission"
draft: false
toc: true
weight: 602
---
## 什么模板可以合并
目前合并进仓库的应用模板，会在「模板市场」中全部展示给用户。
为了控制模板的质量以及避免数量过多带来的繁琐，并不是所有的模板都会被合并到开源仓库中，你可以提前 PR 与我们沟通模板的内容。
预估最后总体的数量不会很多，控制在 50 个左右，一半来自 FastGPT Team，一半来自社区用户。
## 如何写一个应用模板
1. ### 跑通 FastGPT dev 环境
需要在 dev 环境下执行下面的操作。
> 可参照 [FastGPT｜快速开始本地开发](https://doc.fastgpt.in/docs/development/intro/)
1. ### 在 FastGPT 工作台中，创建一个应用
创建空白工作流即可。
![](/imgs/template_submission1.png)
1. ### 创建应用模板
应用模板配置以及相关资源，都会在 **projects/app/public/appMarketTemplates** 目录下。
![](/imgs/template_submission2.png)
1. 在 **projects/app/public/appMarketTemplates** 目录下，创建一个文件夹，名称为模板对应的 id。
2. 在刚刚创建的文件夹中，再创建一个 **template.json** 文件，复制粘贴并填写如下配置：
```JSON
{
"name": "模板名",
"intro": "模板描述，会展示在模板市场的展示页",
"author": "填写你的名字",
"avatar": "模板头像，可以将图片文件放在同一个文件夹中，然后填写相应路径",
"tags": ["模板标签"], // writing(文本创作)，image-generation(图片生成)，web-search(联网搜索),
// roleplay(角色扮演), office-services(办公服务) 暂时分为 5 类，从中选择相应的标签
"type": "模板类别", // simple(简易应用), advanced(工作流), plugin(插件)
"workflow": { // 这个对象先不管，待会直接粘贴导出的工作流即可
"nodes": [],
"edges": [],
"chatConfig": {}
}
}
```
1. ### 完成应用编排并测试
完成应用编排后，可以点击右上角的发布。
1. ### 复制配置到 template.json
鼠标放置在左上角应用的头像和名称上，会出现对于下拉框操作，可以导出工作流配置。
导出的配置，会自动复制到剪切板，可以直接到 template.json 文件中粘贴使用，替换步骤 2 中，**workflow** 的值。
![](/imgs/template_submission3.png)
1. ### 验证模板是否加载成功
刷新页面，打开模板市场，看其是否成功加载，并点击「使用」测试其功能。
![](/imgs/template_submission4.png)
1. ### 提交 PR
如果你觉得你的模板需要提交到开源仓库，可以通过 PR 形式向我们提交。
- 写清楚模板的介绍和功能
- 配上模板运行的效果图
- 模板参数填写说明，需要在 PR 中写清楚。例如，有些模板需要去某个提供商申请 key，需要附上对应的地址和教程，后续我们会加入到文档中。

2217
docSite/content/zh-cn/docs/use-cases/app-cases/translate-subtitle-using-gpt.md Normal file
View File

File diff suppressed because it is too large Load Diff