Chat2Anything

Chat2Anything

| github |

Chat2Anything is a large model (LLM) knowledge base question and answer system for enterprise internal environments. It has the following functions:

• It includes the client (Chat2Anything) and the backend management system (Chat2Anything Admin). The backend management system is responsible for the creation and management of the large model knowledge base. Users can add corporate internal knowledge (Langchain mechanism) to the large model by uploading Excel, Word, PDF and other files. ); the management system includes complete user permissions, roles, organizations, data dictionary management and other functions

• Supports mainstream large models such as ChatGPT, ChatGLM, and Vicuna. Users can switch large models at any time; when using ChatGLM and Vicuna, the entire system can achieve complete private domain deployment, and 100% internal circulation of enterprise data

• Supports business system database access, manual/automatic selection of tables involved in query analysis, and text2SQL based on user problem descriptions (to be improved), and query results are automatically executed and visualized (to be developed)

  
  

deploy

Chat2Anything (client)

1. Locally configure ChatGLM (https://github.com/THUDM/ChatGLM-6B) / Vicuna (https://github.com/lm-sys/FastChat) and other large models, such as using ChatGPT, in Chat2Anything/configs/openai_config. Configure OPENAI_API_KEY and OPENAI_API_BASE in the py file

2. Run controller:

   python3 -m fastchat.serve.controller

3. Run model_worker (ChatGLM-6B, ChatGLM2-6B recommended (https://github.com/THUDM/ChatGLM2-6B)):

   python3 -m fastchat.serve.model_worker_chatglm --model-path /path/to/model/weights

   or

   Run model_worker (vicuna):

   python3 -m fastchat.serve.model_worker --model-path /path/to/model/weights

4. Run client web server

   python3 -m fastchat.serve.chat2anything_web_server

   If using ChatGPT, add the startup parameter --add-chatgpt

5. Basic test environment

   Ubuntu 20.04 CUDA 11.7 Python 3.9 PyTorch 1.13
   Tesla V100 32GB * 1

Chat2Anything Admin (backend management terminal)

1. Database (MySQL) configuration:

    # MySql配置信息
     MYSQL_HOST =数据库地址
     MYSQL_PORT =数据库端口
     MYSQL_DATABASE = chat2anything_db
     MYSQL_USERNAME =账户
     MYSQL_PASSWORD =密码
     
   # 密钥配置
     SECRET_KEY = '一定要改'

2. Model related configuration:

    # model_config.py
     embedding_model_dict = {
     'text2vec": "/Path/to/text2vec-large-chinese' ,
    }
     
   # 密钥配置
     VECTOR_ROOT_PATH = os . path . join ( '/Path/to/Chat2Anything' , 'vector_stores' )

3. Database initialization:

     flask db init
     flask db migrate
     flask db upgrade
     flask admin init

4. run:

     # Windows
     run.bat
   
     # Linux
     chmod +x run.sh
     ./run.sh
   
     # 管理员默写账户密码
     admin 123456

Related projects

Appreciate the following contributions:

• THUDM/chatglm-6b
• lm-sys/FastChat
• THUDM/chatglm2-6b
• chatchat-space/langchain-ChatGLM
• Pear Admin Flask