探索性建模：解锁第三条预训练轴与端到端生成

摘要

由AlexNet开启的深度学习革命告诉我们，端到端训练优于将问题分解为手工设计的阶段。然而，生成建模始终是个例外——尽管生成模型能力卓越，它们仍未实现端到端训练。其根本原因在于，生成建模的核心是处理具有众多模态的分布，而现有可扩展方法处理这一问题的方式相同，即通过分解生成过程，这阻碍了端到端生成。在本工作中，我们提出了探索式建模（Explorative Modeling），一种全新的范式，它转而分解训练循环：在模型生成与数据之间探索K个候选匹配，并基于最优匹配进行训练，从而使预测致力于模态而不是模糊它们。我们发现探索式模型（XMs）在两种场景下具有实用价值。首先，增加探索为现有生成模型在参数和数据之外增添了第三条预训练轴——在此轴上，扩展探索在连续域和离散域（图像、视频和语言）中均能单调地提升性能。值得注意的是，探索带来的收益随规模扩大而增加：随数据规模扩展，收益从7%攀升至36%；随模型规模增大，收益从13%提升至23%；在3倍计算量下，效率增益翻倍以上。具体而言，探索将FLOP效率提升4.1倍，样本效率提升6.2倍，参数效率提升47%，将最强图像生成方案在ImageNet上提升至近乎最先进的1.43 FID（无引导），使现有模型的端到端化程度得以扩展，并解锁了规模化泛化。其次，XMs实现了端到端重构式生成建模，在控制任务上以16-256倍更少的推理步数即可媲美扩散模型。综合这些结果，XMs既确立了作为现有生成模型新预训练轴的地位，也确立了一种独立的端到端生成建模范式。

English

The deep learning revolution, kicked off by AlexNet, taught us that end-to-end training beats decomposing a problem into hand-designed stages. Generative modeling, however, has remained the exception-despite generative models being remarkably capable, they are still not trained end-to-end. This is because, at its core, generative modeling is about handling distributions with many modes, and existing scalable approaches handle this the same way, by factoring the generation procedure, which prevents end-to-end generation. In this work, we introduce Explorative Modeling, a new paradigm that instead factors the training loop, exploring K candidate matches between model generations and data, and training on the best, so predictions commit to modes rather than blurring them. We find Explorative Models (XMs) useful in two settings. First, increasing exploration adds a third pretraining axis beyond parameters and data for existing generative models-where scaling exploration monotonically improves performance across both continuous and discrete domains (images, video, and language). Notably, gains from exploration increase with scale, climbing from 7% to 36% as data scales and from 13% to 23% as models grow, with efficiency gains more than doubling at 3x the compute. Concretely, exploration improves FLOP efficiency by 4.1x, sample efficiency by 6.2x, parameter efficiency by 47%, lifts the strongest of image-generation recipes to a near-state-of-the-art 1.43 FID on ImageNet without guidance, enables scaling how end-to-end existing models are, and unlocks scaling generalization. Second, XMs enable end-to-end reconstructive generative modeling, matching diffusion on control tasks with 16-256x fewer inference steps. Together, these results establish XMs as both a new pretraining axis for existing generative models and a standalone end-to-end generative modeling paradigm.