Bigger is not Always Better: Scaling Properties of Latent Diffusion Models
This work addresses the problem of optimizing generative model efficiency for practitioners, offering insights that could lead to more cost-effective scaling strategies, though it is incremental as it builds on existing diffusion model research.
The paper investigates how model size affects the sampling efficiency of latent diffusion models, finding that smaller models often produce higher-quality results than larger ones under the same inference budget.
We study the scaling properties of latent diffusion models (LDMs) with an emphasis on their sampling efficiency. While improved network architecture and inference algorithms have shown to effectively boost sampling efficiency of diffusion models, the role of model size -- a critical determinant of sampling efficiency -- has not been thoroughly examined. Through empirical analysis of established text-to-image diffusion models, we conduct an in-depth investigation into how model size influences sampling efficiency across varying sampling steps. Our findings unveil a surprising trend: when operating under a given inference budget, smaller models frequently outperform their larger equivalents in generating high-quality results. Moreover, we extend our study to demonstrate the generalizability of the these findings by applying various diffusion samplers, exploring diverse downstream tasks, evaluating post-distilled models, as well as comparing performance relative to training compute. These findings open up new pathways for the development of LDM scaling strategies which can be employed to enhance generative capabilities within limited inference budgets.