The Test-Time Compute Scaling: Smarter LLMs Without Retraining video course shows you how to get dramatically better answers from the models you already have, without touching a single weight. You'll gain a real competitive advantage, sharper judgment about where to invest compute, and skills that keep you adaptable as the field keeps evolving. If you've ever felt like your model knows more than it's telling you, this course is where you learn to draw it out.

This certification covers the following topics:

  • The Large Language Monkeys: Sampling Your Way to Smarter Answers
  • Inference Has Scaling Laws Too
  • The Long Tail of Hard Problems
  • The Verifier Problem: Generation Is Easy, Verification Is Hard
  • Reward Models: Teaching Machines to Judge
  • Sequential Revisions: The Other Dimension of Scaling
  • Test-Time Compute vs. Pre-Training: Where Should You Spend Your Tokens?
  • The Building Blocks: Seven Operations That Power Inference Architectures
  • Archon: Treating Inference as an Architecture Search Problem