Slim

Slim - Lightweight client for local LLM inference

Provides a simple interface to Ollama for running small language models. Designed for specialized tasks like classification, annotation, and extraction.

Usage

require "slim"

client = Slim::Client.new

# Simple completion
response = client.generate("llama3.2:3b", "Classify this as question or statement: How are you?")
puts response.text

# Streaming
client.generate("llama3.2:3b", prompt) do |chunk|
  print chunk
end

# Chat format
messages = [
  Slim::Message.new("system", "You are a classifier."),
  Slim::Message.new("user", "Is this a question? Hello world"),
]
response = client.chat("llama3.2:3b", messages)