Running LLMs Locally with Ollama: Privacy and Performance

curl -fsSL https://ollama.ai/install.sh | sh

brew install ollama

ollama --version

# Install Llama 2 7B
ollama pull llama2

# Install Code Llama for programming
ollama pull codellama

# Install Mistral 7B
ollama pull mistral

ollama run llama2

ollama run llama2 "Explain what DevOps is"

curl http://localhost:11434/api/generate -d '{
  "model": "llama2",
  "prompt": "Why use Kubernetes?",
  "stream": false
}'

import requests
import json

def query_ollama(prompt, model="llama2"):
    url = "http://localhost:11434/api/generate"
    data = {
        "model": model,
        "prompt": prompt,
        "stream": False
    }
    
    response = requests.post(url, json=data)
    return json.loads(response.text)["response"]

# Usage
result = query_ollama("How to implement CI/CD?")
print(result)

const axios = require('axios');

async function queryOllama(prompt, model = 'llama2') {
    try {
        const response = await axios.post('http://localhost:11434/api/generate', {
            model: model,
            prompt: prompt,
            stream: false
        });
        
        return response.data.response;
    } catch (error) {
        console.error('Error:', error);
    }
}

// Usage
queryOllama('Explain Docker containers')
    .then(result => console.log(result));

FROM llama2

# Set temperature (creativity)
PARAMETER temperature 0.7

# Set system prompt
SYSTEM """
You are a specialist in DevOps and Cloud Computing.
Always respond in a technical and practical manner.
"""

ollama create devops-expert -f ./Modelfile
ollama run devops-expert

# Set number of threads
export OLLAMA_NUM_THREADS=8

# Use GPU (if available)
export OLLAMA_GPU=1

# Configure memory
export OLLAMA_MAX_LOADED_MODELS=2

# View loaded models
ollama list

# Monitor resource usage
htop
nvidia-smi  # For GPU

ollama run codellama "Create a Python function to validate a CPF"

ollama run llama2 "Analyze this error: $(cat error.log)"

ollama run codellama "Document this function: $(cat function.py)"

ollama run codellama "Review this code and suggest improvements: $(cat script.sh)"

# Check disk space
df -h

# Check available RAM
free -h

# Ollama logs
ollama logs

# Check if GPU is being used
nvidia-smi

# Adjust threads
export OLLAMA_NUM_THREADS=4

# Check if the service is running
ps aux | grep ollama

# Restart service
ollama serve