Choosing an Alignment Method for a Resource-Constrained Project

Which of the following best analyzes the primary reason why Direct Policy Optimization (DPO) is considered more sample-efficient than Proximal Policy Optimization (PPO) for aligning language models?

The primary reason Direct Policy Optimization (DPO) is considered more sample-efficient than Proximal Policy Optimization (PPO) is that DPO requires actively collecting new preference data from an online environment throughout its training process.