Generative models form the backbone of modern machine learning, underpinning state-of-the-art systems in text, vision, and multimodal applications. While Maximum Likelihood Estimation has traditionally served as the dominant training paradigm, recent work have highlighted its limitations, particularly in generalization and susceptibility to catastrophic forgetting compared to Reinforcement Learning techniques, such as Policy Gradient methods. However, these approaches depend on explicit reward signals, which are often unavailable in practice, leaving open the fundamental problem of how to align generative models when only high-quality datasets are accessible. In this work, we address this challenge via a Bilevel Optimization framework, where the reward function is treated as the optimization variable of an outer-level problem, while a policy gradient objective defines the inner-level. We then conduct a theoretical analysis of this optimization problem in a tractable setting and extract insights that, as we demonstrate, generalize to applications such as tabular classification and model-based reinforcement learning. We release the code at https://github.com/abenechehab/nll_to_po .

翻译：生成模型构成了现代机器学习的支柱，支撑着文本、视觉和多模态应用中最先进的系统。尽管最大似然估计传统上作为主导的训练范式，但近期研究凸显了其局限性，特别是在泛化能力和对灾难性遗忘的易感性方面，与强化学习技术（如策略梯度方法）相比尤为明显。然而，这些方法依赖于显式的奖励信号，而这些信号在实践中往往难以获取，这留下了一个根本性问题：当仅能获取高质量数据集时，如何对齐生成模型？在本研究中，我们通过双层优化框架应对这一挑战，其中奖励函数被视为外层问题的优化变量，而策略梯度目标则定义了内层问题。随后，我们在一个可处理的设定下对该优化问题进行了理论分析，并提取了具有普适性的洞见——正如我们通过实验所展示的，这些洞见可推广至表格分类和基于模型的强化学习等应用。代码发布于 https://github.com/abenechehab/nll_to_po。