Turning a frontier model's judgment into training signal for your open model, without crossing the line its terms of service draws.
Part of: Fine-Tuning & Distilling Open Models