ollama

Overview

The ollama node runs open-source LLMs locally using Ollama. Perfect for edge computing where privacy is critical, internet is unavailable, or you need cost-free unlimited inference. Supports Llama, Mistral, CodeLlama, and more.

Properties

Popular Models

Installing Ollama

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

brew install ollama

docker run -d -v ollama:/root/.ollama -p 11434:11434 ollama/ollama

Example: Local Sensor Classification

Classify sensor readings locally without sending data to the cloud.

// Function node: Classify sensor data locally
var reading = msg.payload;

msg.systemPrompt = `You are a sensor data classifier. Classify readings into:
- NORMAL: Within expected range
- WARNING: Approaching limits
- CRITICAL: Requires immediate attention

Respond with ONLY the classification and a brief reason.`;

msg.payload = `Classify this sensor reading:
Temperature: ${reading.temperature}°C
Humidity: ${reading.humidity}%
Pressure: ${reading.pressure} hPa

Normal ranges:
- Temperature: 18-26°C
- Humidity: 30-60%
- Pressure: 980-1020 hPa`;

msg.temperature = 0.1; // Very deterministic
msg.model = "llama3.2"; // Fast, small model

return msg;

Example: Local Image Analysis with LLaVA

Analyze camera images locally using the LLaVA vision model.

// Function node: Analyze image locally
var imageBuffer = msg.payload;
var base64Image = imageBuffer.toString('base64');

msg.model = "llava";
msg.payload = "Describe what you see in this image. Focus on any people or vehicles.";
msg.images = [base64Image];

return msg;

// Output processing
// Function node: Parse and act on results
if (msg.payload.toLowerCase().includes("person")) {
    msg.alert = true;
    msg.alertType = "motion_detected";
}
return msg;

Performance Tips