强化学习

📄️Self-Improving Vision-Language-Action Models with Data Generation via Residual RL

PLD 方法整体流程