Chat-Anything-Robot

CAR：A chat robot with any fields of given knowledge

Quick Start

Create Environment

pip+virtualenv

• Clone the project repository:

  git clone https://github.com/Bili-Sakura/Chat-Anything-Robot.git
  cd Chat-Anything-Robot

• Create and activate virtual environment:

  Recommended python version is 3.8.x

  python -m venv .venv/car  # Create a virtual environment named car, located in .venv
  source .venv/car/bin/activate  # For Unix-like systems
  .\.venv\car\Scripts\activate  # For Windows

• Install dependencies:

  pip install -r requirements.txt

• Deactivate the virtual environment:

  source .venv/car/bin/deactivate  # For Unix-like systems
  .\.venv\car\Scripts\deactivate # For Windows

conda

• Create and activate conda environment:

  conda create --name car python=3.8
  conda activate car

• Install dependencies:

  conda install --file requirements.txt

• Deactivate the conda environment:

  conda deactivate

Prepare OpenAI API KEY

Follow the steps below to prepare your OpenAI API key:

1. Sign up for an account on OpenAI's website.
2. Generate an API key in your account settings.
3. Copy the API key to use in your application.

It is avaliable to use api key transfered from supported website. Set base url parameter whenever you use.

Run

cd Chat-Anything-Robot
run.bat # which equals to [cmd].venv\car\Scripts\activate + [cmd]streamlit run app.py

Congratulations! You've start your own program!

Add Knowledge to Your Customed LLM

Put files (*.docx, *.txt, et al.) under the data folder.

Re-init the program, the corpus will be added into vectorstore automatically (may cost time in the first time).

Development Notes

Feature

• Add button for vercterization
• Add cost log info locally
• Add knowledge info in sidebar

Fix

• Asynchronously chat

Refactor

• Web Application Project Structure