Microsoft’s rStar2-Agent: How a 14B Model Learned to “Think Smarter” and Out-Reason AI Giants

​ A novel RL algorithm, a hyper-efficient infrastructure, and a counter-intuitive training recipe — redefining the frontier of AI reasoning.Continue reading on Medium »   Read More AI on Medium 

#AI