September 25, 2025
2025
New work “RL Grokking Recipe: How Does RL Unlock New Algorithms in LLMs?” [Code & Data]