github/DB-GPT
1
0
Fork 0
mirror of https://github.com/csunny/DB-GPT.git synced 2025-07-29 23:01:38 +00:00
Code Issues Packages Projects Releases Wiki Activity

Update README.md

Browse Source
This commit is contained in:
rinne1998 2023-05-16 18:03:18 +08:00
parent ae947f2310
commit f20bf41278
2 changed files with 159 additions and 55 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

177
README.en.md
View File

@ -8,80 +8,163 @@ A Open Database-GPT Experiment, interact your data and environment using the loc
## Features
- SQL Project
- SQL Generate
- SQL-diagnosis
- Database-QA Based Knowledge
We have currently released various features, which are listed below to showcase the capabilities that have been released so far:
- SQL language capabilities
- SQL generation
- SQL diagnosis
- Private domain Q&A and data processing
- Database knowledge Q&A
- Data processing
- Plugin model
- Supports custom plugins to perform tasks, with native support for Auto-GPT plugins. For example:
- Automatic SQL execution to obtain query results
- Automatic crawling of learning knowledge
- Unified vector storage/indexing of knowledge base
- Support for unstructured data
- PDF, Markdown, CSV, WebURL
## Architecture Design
<p align="center">
<img src="./assets/DB-GPT.png" width="600px" />
</p>
[DB-GPT](https://github.com/csunny/DB-GPT) is an experimental open-source application that builds upon the [FastChat](https://github.com/lm-sys/FastChat) model and uses vicuna as its base model. Additionally, it looks like this application incorporates langchain and llama-index embedding knowledge to improve Database-QA capabilities.
Overall, it appears to be a sophisticated and innovative tool for working with databases. If you have any specific questions about how to use or implement DB-GPT in your work, please let me know and I'll do my best to assist you.
## Demo
Run on an RTX 4090 GPU (The origin mov not sped up!, [YouTube地址](https://www.youtube.com/watch?v=1PWI6F89LPo))
Run on an RTX 4090 GPU. [YouTube](https://www.youtube.com/watch?v=1PWI6F89LPo)
### Run
<p align="center">
<img src="./assets/演示.gif" width="600px" />
</p>
### SQL Generate
First, select the DataBase, you can use Schema to generate the SQL.。
<p align="center">
<img src="./assets/SQLGEN.png" width="600px" />
<img src="./assets/Auto-DB-GPT.gif" width="600px" />
</p>
### SQL Generate
1. Generate SQL based on schema information.
<p align="center">
<img src="./assets/SQL_Gen_CreateTable.png" width="600px" />
</p>
2. Generate executable SQL.
First, select the corresponding database, and then the model can generate SQL based on the corresponding database schema information. The successful execution result is demonstrated below:
<p align="center">
<img src="./assets/exeable.png" width="600px" />
</p>
### Database-QA
### Q&A
<p align="center">
<img src="./assets/DB_QA.png" width="600px" />
</p>
1. Based on the default built-in knowledge base, question and answer.
<p align="center">
<img src="./assets/VectorDBQA.png" width="600px" />
</p>
2. Add your own knowledge base.
<p align="center">
<img src="./assets/new_knownledge.gif" width="600px" />
</p>
3. Learn by scraping data from the internet.
- TODO
## Architecture Design
[DB-GPT](https://github.com/csunny/DB-GPT) is an experimental open-source application that builds upon the [FastChat](https://github.com/lm-sys/FastChat) model and uses vicuna as its base model. Additionally, it looks like this application incorporates langchain and llama-index embedding knowledge to improve Database-QA capabilities.
Overall, it appears to be a sophisticated and innovative tool for working with databases. If you have any specific questions about how to use or implement DB-GPT in your work, please let me know and I'll do my best to assist you.
<p align="center">
<img src="./assets/DB-GPT.png" width="600px" />
</p>
The core capabilities mainly include the following parts:
1. Knowledge base
2. LLMs management
3. Vector storage and indexing
4. Connections
5. Agent and plugins
6. Prompt generation and optimization
7. Multi-end product interface
Below is a brief introduction to each module:
### Knowledge Base
As the most important scenario for current user needs, we natively support the construction and processing of knowledge bases. At the same time, in this project, we also provide various knowledge base management strategies, such as:
1. Default built-in knowledge base
2. Customized new knowledge base
3. Building knowledge bases through plug-in capabilities and other usage scenarios.
Users only need to organize their knowledge documents and use our existing capabilities to build the knowledge base required for large models.
### LLMs Management
In the underlying large model integration, we have designed an open interface that supports integration with various large models. At the same time, we have a very strict control and evaluation mechanism for the effectiveness of the integrated models. In terms of accuracy, the integrated models need to align with the capability of ChatGPT at a level of 85% or higher. We use higher standards to select models, hoping to save users the cumbersome testing and evaluation process in the process of use.
### Vector storage and indexing
In order to facilitate the management of knowledge after vectorization, we have built-in multiple vector storage engines, from memory-based Chroma to distributed Milvus. Users can choose different storage engines according to their own scenario needs. The storage of knowledge vectors is the cornerstone of AI capability enhancement. As the intermediate language for interaction between humans and large language models, vectors play a very important role in this project.
### Connections
In order to interact more conveniently with users' private environments, the project has designed a connection module, which can support connection to databases, Excel, knowledge bases, and other environments to achieve information and data exchange.
### Agent and Plugin
The ability of Agent and Plugin is the core of whether large models can be automated. In this project, we natively support the plugin mode, and large models can automatically achieve their goals. At the same time, in order to give full play to the advantages of the community, the plugins used in this project natively support the Auto-GPT plugin ecology, that is, Auto-GPT plugins can directly run in our project.
### Prompt Automatic Generation and Optimization
Prompt is a very important part of the interaction between the large model and the user, and to a certain extent, it determines the quality and accuracy of the answer generated by the large model. In this project, we will automatically optimize the corresponding prompt according to user input and usage scenarios, making it easier and more efficient for users to use large language models.
### Multi-Platform Product Interface
TODO: In terms of terminal display, we will provide a multi-platform product interface, including PC, mobile phone, command line, Slack and other platforms.
## Deployment
### 1. Python Requirement
### 1. Hardware Requirements
As our project has the ability to achieve ChatGPT performance of over 85%, there are certain hardware requirements. However, overall, the project can be deployed and used on consumer-grade graphics cards. The specific hardware requirements for deployment are as follows:
```bash
$ python>=3.9
$ pip install -r requirements.txt
```
| GPU | VRAM Size | Performance |
| --------- | --------- | ------------------------------------------- |
| RTX 4090 | 24 GB | Smooth conversation inference |
| RTX 3090 | 24 GB | Smooth conversation inference, better than V100 |
| V100 | 16 GB | Conversation inference possible, noticeable stutter |
### 2. Install
or if you use conda envirenment, you can use this command
```bash
$ conda env create -f environment.yml
```
### 2. MySQL
In this project examples, we connect mysql and run SQL-Generate. so you need install mysql local for test. recommand docker
This project relies on a local MySQL database service, which you need to install locally. We recommend using Docker for installation.
```bash
$ docker run --name=mysql -p 3306:3306 -e MYSQL_ROOT_PASSWORD=aa12345678 -dit mysql:latest
```
We use [Chroma embedding database](https://github.com/chroma-core/chroma) as the default for our vector database, so there is no need for special installation. If you choose to connect to other databases, you can follow our tutorial for installation and configuration.
For the entire installation process of DB-GPT, we use the miniconda3 virtual environment. Create a virtual environment and install the Python dependencies.
```
python>=3.10
conda create -n dbgpt_env python=3.10
conda activate dbgpt_env
pip install -r requirements.txt
```
Alternatively, you can use the following command:
```
cd DB-GPT
conda env create -f environment.yml
```
It is recommended to set the Python package path to avoid runtime errors due to package not found.
```
echo "/root/workspace/DB-GPT" > /root/miniconda3/env/dbgpt_env/lib/python3.10/site-packages/dbgpt.pth
```
### 3. LLM
- [vicuna](https://github.com/lm-sys/FastChat/blob/main/README.md#model-weights)
- [Hugging Face](https://huggingface.co/Tribbiani/vicuna-7b)
### 3. Run
You can refer to this document to obtain the Vicuna weights: [Vicuna](https://github.com/lm-sys/FastChat/blob/main/README.md#model-weights)
If you have difficulty with this step, you can also directly use the model from [this link](https://huggingface.co/Tribbiani/vicuna-7b) as a replacement.
```bash
$ cd pilot/server
@ -94,6 +177,12 @@ Run gradio webui
$ python webserver.py
```
## Usage Instructions
We provide a user interface for Gradio, which allows you to use DB-GPT through our user interface. Additionally, we have prepared several reference articles (written in Chinese) that introduce the code and principles related to our project.
1. [大模型实战系列(1) —— 强强联合Langchain-Vicuna应用实战](https://zhuanlan.zhihu.com/p/628750042)
2. [大模型实战系列(2) —— DB-GPT 阿里云部署指南](https://zhuanlan.zhihu.com/p/629467580)
3. [大模型实战系列(3) —— DB-GPT插件模型原理与使用](https://zhuanlan.zhihu.com/p/629623125)
## Thanks
- [FastChat](https://github.com/lm-sys/FastChat)
@ -101,6 +190,18 @@ $ python webserver.py
- [langchain](https://github.com/hwchase17/langchain)
- [llama-index](https://github.com/jerryjliu/llama_index) and [In-Context Learning](https://arxiv.org/abs/2301.00234)
<!-- GITCONTRIBUTOR_START -->
## Contributors
|[<img src="https://avatars.githubusercontent.com/u/17919400?v=4" width="100px;"/><br/><sub><b>csunny</b></sub>](https://github.com/csunny)<br/>|[<img src="https://avatars.githubusercontent.com/u/1011681?v=4" width="100px;"/><br/><sub><b>xudafeng</b></sub>](https://github.com/xudafeng)<br/>|[<img src="https://avatars.githubusercontent.com/u/7636723?s=96&v=4" width="100px;"/><br/><sub><b>明天</b></sub>](https://github.com/yhjun1026)<br/> | [<img src="https://avatars.githubusercontent.com/u/13723926?v=4" width="100px;"/><br/><sub><b>Aries-ckt</b></sub>](https://github.com/Aries-ckt)<br/>|[<img src="https://avatars.githubusercontent.com/u/95130644?v=4" width="100px;"/><br/><sub><b>thebigbone</b></sub>](https://github.com/thebigbone)<br/>|
| :---: | :---: | :---: | :---: |:---: |
This project follows the git-contributor [spec](https://github.com/xudafeng/git-contributor), auto updated at `Sun May 14 2023 23:02:43 GMT+0800`.
<!-- GITCONTRIBUTOR_END -->
## Licence
The MIT License (MIT)

37
README.md
View File

@ -24,7 +24,7 @@ DB-GPT 是一个开源的以数据库为基础的GPT实验项目，使用本地
- 自动爬取学习知识
- 知识库统一向量存储/索引
- 非结构化数据支持
- PDF、MarkDown、CSV、WebURL
- PDF、Markdown、CSV、WebURL
## 效果演示
@ -76,7 +76,7 @@ DB-GPT 是一个开源的以数据库为基础的GPT实验项目，使用本地
- TODO
## 架构方案
DB-GPT基于[FastChat](https://github.com/lm-sys/FastChat) 构建大模型运行环境，并提供 vicuna 作为基础的大语言模型。此外，我们通过langchain提供私域知识库问答能力。同时我们支持插件模式, 在设计上原生支持Auto-GPT插件。
DB-GPT基于 [FastChat](https://github.com/lm-sys/FastChat) 构建大模型运行环境，并提供 vicuna 作为基础的大语言模型。此外，我们通过LangChain提供私域知识库问答能力。同时我们支持插件模式, 在设计上原生支持Auto-GPT插件。
整个DB-GPT的架构，如下图所示
@ -96,7 +96,12 @@ DB-GPT基于[FastChat](https://github.com/lm-sys/FastChat) 构建大模型运行
下面对每个模块也做一些简要的介绍:
### 知识库能力
知识库作为当前用户需求最大的场景，我们原生支持知识库的构建与处理。同时在本项目当中，也提供了多种知识库的管理策略。 如: 1. 默认内置知识库 2. 自定义新增知识库 3. 通过插件能力自抓取构建知识库等多种使用场景。 用户只需要整理好知识文档，即可用我们现有的能力构建大模型所需要的知识库能力。
知识库作为当前用户需求最大的场景，我们原生支持知识库的构建与处理。同时在本项目当中，也提供了多种知识库的管理策略。 如:
1. 默认内置知识库
2. 自定义新增知识库
3. 通过插件能力自抓取构建知识库等多种使用场景。
用户只需要整理好知识文档，即可用我们现有的能力构建大模型所需要的知识库能力。
### 大模型管理能力
在底层大模型接入中，设计了开放的接口，支持对接多种大模型。同时对于接入模型的效果，我们有非常严格的把控与评审机制。对大模型能力上与ChatGPT对比，在准确率上需要满足85%以上的能力对齐。我们用更高的标准筛选模型，是期望在用户使用过程中，可以省去前面繁琐的测试评估环节。
@ -114,20 +119,18 @@ Agent与插件能力是大模型能否自动化的核心，在本的项目中，
Prompt是与大模型交互过程中非常重要的部分，一定程度上Prompt决定了大模型生成答案的质量与准确性，在本的项目中，我们会根据用户输入与使用场景，自动优化对应的Prompt，让用户使用大语言模型变得更简单、更高效。
### 多端产品界面
TODO: 在终端展示上，我们将提供多端产品界面。包括PC、手机、命令行、slack等多种模式。
TODO: 在终端展示上，我们将提供多端产品界面。包括PC、手机、命令行、Slack等多种模式。
## 安装教程
### 硬件说明
### 1.硬件说明
因为我们的项目在效果上具备ChatGPT 85%以上的能力，因此对硬件有一定的要求。 但总体来说，我们在消费级的显卡上即可完成项目的部署使用，具体部署的硬件说明如下:
```
GPU型号 | 显存大小 | 性能
-------|----------|------------------------------
TRX4090| 24G |可以流畅的进行对话推理，无卡顿
TRX3090| 24G |可以流畅进行对话推理，有卡顿感，但好与V100
V100 | 16G |可以进行对话推理，有明显卡顿
```
### DB-GPT安装
| GPU型号 | 显存大小 | 性能 |
| ------- | -------- | ------------------------------------------ |
| RTX4090 | 24G | 可以流畅的进行对话推理，无卡顿 |
| RTX3090 | 24G | 可以流畅进行对话推理，有卡顿感，但好于V100 |
| V100 | 16G | 可以进行对话推理，有明显卡顿 |
### 2.DB-GPT安装
本项目依赖一个本地的 MySQL 数据库服务，你需要本地安装，推荐直接使用 Docker 安装。
```
@ -154,10 +157,10 @@ echo "/root/workspace/DB-GPT" > /root/miniconda3/env/dbgpt_env/lib/python3.10/si
### 3. 运行大模型
关于基础模型, 可以根据[vicuna](https://github.com/lm-sys/FastChat/blob/main/README.md#model-weights)合成教程进行合成。
如果此步有困难的同学，也可以直接使用[Hugging Face](https://huggingface.co/)上的模型进行替代. [替代模型](https://huggingface.co/Tribbiani/vicuna-7b)
关于基础模型, 可以根据[Vicuna](https://github.com/lm-sys/FastChat/blob/main/README.md#model-weights)合成教程进行合成。
如果此步有困难的同学，也可以直接使用[此链接](https://huggingface.co/Tribbiani/vicuna-7b)上的模型进行替代。
2. 运行模型服务
运行模型服务
```
cd pilot/server
python llmserver.py
@ -171,7 +174,7 @@ $ python webserver.py
## 使用说明
我们提供了gradio的用户界面，可以通过我们的用户界面使用DB-GPT， 同时关于我们项目相关的一些代码跟原理介绍，我们也准备了以下几篇参考文章。
我们提供了Gradio的用户界面，可以通过我们的用户界面使用DB-GPT， 同时关于我们项目相关的一些代码跟原理介绍，我们也准备了以下几篇参考文章。
1. [大模型实战系列(1) —— 强强联合Langchain-Vicuna应用实战](https://zhuanlan.zhihu.com/p/628750042)
2. [大模型实战系列(2) —— DB-GPT 阿里云部署指南](https://zhuanlan.zhihu.com/p/629467580)
3. [大模型实战系列(3) —— DB-GPT插件模型原理与使用](https://zhuanlan.zhihu.com/p/629623125)