Russ Tedrake, MIT and Toyota Research Institute: "Large Behavior Models for Dexterous Manipulation"

New short course: Reinforcement Fine-Tuning LLMs with GRPO! Reasoning models have been one of the most important developments in LLMs. Learn how to train LLMs for complex reasoning tasks, like solving math problems, generating code, or playing Wordle, without relying on large labeled datasets. In this course made in collaboration with Predibase HQ, Travis Addair and Arnav Garg teach you how to use GRPO, a reinforcement learning algorithm that guides models using reward functions instead of human