TL;DR

Run Kubernetes Deployment & Simulation:

% ./simulation/create.sh
% curl http://localhost:30000/hello-app

# Get istio proxy logs for hello-app
% kubectl logs -c istio-proxy --selector app=hello-app
# Get container logs for hello-app
% kubectl logs--selector app=hello-app

# Tear it down
% ./simulation/destroy.sh

Run ollama app:

% pip install BeautifulSoup4 langchain langchain-community langchain-core tikToken chromadb
% ollama pull nomic-embed-text
% ollama pull mistral
% python app/app.py

What Does it Do?

We want to demonstrate how to ask the Llama LLM a question, demonstrate how to fetch data about the question, feed data as context, ask the exact same question again and receive different (better) answer.

1. Ask llama: What is Ollama? Receive response X.
2. Pull text from 3 web pages, create embeddings from model, and store in Chroma vector db: https://ollama.com, https://ollama.com/blog/windows-preview, https://ollama.com/blog/openai-compatibility
3. Re-Ask llama the same question but add context data from step 2: What is Ollama? Receive response Y.

Overview

This project is a playground for tinkering with llama3 genai.

Technologies

• llama3
• ollama (https://github.com/ollama/ollama)

Links

• Starting point gathered from https://www.youtube.com/watch?v=jENqvjpkwmw&ab_channel=MervinPraison