SFT or RL? An Early Investigation into Training R1-Like Reasoning Large Vision-Language Models Paper • 2504.11468 • Published 15 days ago • 27 • 2