This project is a tool for analyzing the thought process of LLMs. It uses a combination of techniques to extract the thought process from the LLM's output and visualize it in a graph.
pip install -r requirements.txt
pip install -e .
And run the following command to start the server:
streamlit run app.py
We've prepared a few sample files to test the app. You can find them in the experimentsd_data
folder.
The app will load the JSON file and display the graph. You can then select a node to see the details of the node.