Can a 1.7B Model “Think Like” a 20B One? I Tried the Advisor Pattern with Ollama

1. 1Implement the Advisor Tool from Anthropic to leverage a cheaper model for most tasks
2. 2Use the Advisor Pattern to 'phone a smarter friend' only when necessary
3. 3Configure the executor to run the loop and the advisor to steer when needed
4. 4Test and evaluate the performance of the Advisor Pattern with Ollama
5. 5Apply the Advisor Pattern to other models and tasks to improve efficiency and reduce costs

ML engineers and researchers can benefit from this technique to optimize model performance and reduce costs

💡 The Advisor Pattern can significantly improve model efficiency by leveraging a cheaper model for most tasks and only using a smarter model when necessary

🤖 Make a 1.7B model think like a 20B one with the Advisor Pattern! 🚀