Skip to content

Latest commit

 

History

History
36 lines (26 loc) · 813 Bytes

README.md

File metadata and controls

36 lines (26 loc) · 813 Bytes

KG_vs_Agent

image

Talk Lyon Data Science on comparing a Knowledge Graph approach to a Web Agent one. Replay of the talk can be found on YouTube, the talk is in French.

KG-RAG

Initialise the project:

git clone https://github.com/Lyon-NLP/KG_vs_Agent.git
cd KG_vs_Agent
cd kg_rag
python3 -m venv kg_env
source kg_env/bin/active
pip install -r requirements.txt

Start the conversational interface on Streamlit:

streamlit run main.py

Specify an .env file with :

  • the LLM model name
  • the LLM API key
  • the Neo4j graph URL
  • the Neo4j graph user
  • the Neo4j graph password
  • the Neo4j graph database name

Web Agent

Please check the REDAME in the web_agent folder.