About

This is the code repository accompanying the Medium Article: A Starter Pack to building a local Chatbot using Ollama, LangGraph and RAG

Setup

Ollama

1. Download Ollama You can download Ollama from https://www.ollama.com.

2. Verify Ollama is Running

• Once installed, ensure that Ollama is running by accessing: http://localhost:11434/

3. Pull Required Models

• With Ollama running, you’ll need to pull the following three models:

ollama pull nomic-embed-text
ollama pull llama3.1
ollama pull deepseek-r1:8b

╔══════════════════╦═════════════════════════════════════════╗
║       Name       ║                  Usage                  ║
╠══════════════════╬═════════════════════════════════════════╣
║ nomic-embed-text ║ text embedding for RAG                  ║
║ llama3.1         ║ Simple Chat, Fast Response              ║
║ deepseek-r1:8b   ║ Complex Chat, Well thought out Response ║
╚══════════════════╩═════════════════════════════════════════╝

Python environment

conda env create -f env.yml
conda activate chatbot

Create vector store

python create_vs.py

Run webapp

streamlit run app.py

Questions you can ask

• Trigger RAG: ask question related to Kredivo
• Trigger system 2: ask it to plan an itinerary