it's learning how to evolve," commented Shashank Yadav, CEO of Fraction AI. "QwQ-32B proves that reinforcement learning can ...