A toy demo to play with langchain.

It can generate and execute commands (in docker container for safety) from user's natural language.

And it rely on ollama to chat with locally deployed LLM (qwen-14b).

Setup

1. Install packages

pip install -r requirements.txt

2. Install and serve docker, ollama
3. Pull the docker image ubuntu:22.04 and LLM qwen:14b

Usage

Start chat session:

python main.py

Special input seq:

• clean/clear: Clear chat history of current session
• exit/quit/q: Exit chat session