https://github.com/tongjilibo/chat4doc

RAG的最小实现
https://github.com/tongjilibo/chat4doc

bert4torch llm rag

Last synced: 9 months ago
JSON representation

RAG的最小实现

• Host: GitHub
• URL: https://github.com/tongjilibo/chat4doc
• Owner: Tongjilibo
• License: apache-2.0
• Created: 2023-12-10T12:26:55.000Z (over 2 years ago)
• Default Branch: master
• Last Pushed: 2024-01-02T14:59:38.000Z (over 2 years ago)
• Last Synced: 2025-08-02T16:44:18.616Z (10 months ago)
• Topics: bert4torch, llm, rag
• Language: CSS
• Homepage:
• Size: 1.21 MB
• Stars: 4
• Watchers: 1
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

README

          
# chat4doc

RAG的最小实现

![chat4doc](./doc/example.png)

## Quick start

- 配置config文件, 若修改需同步修改`client_py/static/js/script.js`中相关配置

- 执行`sh start.sh`开启服务

- 访问[http://localhost:8083/chat4doc](http://localhost:8083/chat4doc)，并上传pdf进行解析

## 文件说明

- `client_py`: 前端webui

- `server`: 后端（含llm和召回模块）

## Todo

- ~~文章加载完先调用大模型简单介绍pdf的情况~~

- 流式回复

- 出处的地方可以点击，并在pdf上跳转到指定的页面并高亮

- pdf展示这块使用html的格式

- 提升召回模块的准确性

- 适配更多的大模型

## reference

- [chatpdf](https://github.com/shibing624/ChatPDF)

- [ChatDocs](https://github.com/marella/chatdocs)

- [DocsGPT](https://github.com/arc53/DocsGPT)

- [ChatLLM](https://github.com/yuanjie-ai/ChatLLM)

- [pdfGPT](https://github.com/bhaskatripathi/pdfGPT)

- [gpt_academic](https://github.com/binary-husky/gpt_academic)

- [ChatPaper](https://github.com/kaixindelele/ChatPaper)

- [chatdoc](https://github.com/daodao97/chatdoc)