This repository that demonstrates Chain-of-Table reasoning on multiple tables, which is SOTA research by folks at Google and Stanford powered by LangGraph.
Follow these steps to set up:
- Clone the Repository
git clone https://github.com/aaliyan1230/chain-of-table-langgraph.git
- Configure environment
Navigate to the repository directory and run:
python -m venv venv
source venv/bin/activate
- Install Dependencies
Navigate to the repository directory and run:
pip install -r requirements.txt
- Configure API Keys
Create a
.env
file in the root directory. Add your OpenAI API key and LangChain API details as follows, if you don't require logging via langsmith, only openai key would do fine too:OPENAI_API_KEY="..." LANGCHAIN_API_KEY="..." LANGCHAIN_TRACING_V2="..." LANGCHAIN_ENDPOINT="..." LANGCHAIN_PROJECT="..."
run the main.py as a python script to run the implementation in a terminal
python chain_of_table/main.py
present inside the data/ directory