In recent years, a significant focus of research has been on architected materials, including mechanical metamaterials. The increased prevalence of additive manufacturing technologies and generative design methodologies has led to demonstrable industrial relevance in a wide range of applications, including sports equipment, ballistic protection, and medical implants [1]. These applications require distinct responses at different strain rates [2]. A majority of existing studies have concentrated on the behavior of single unit cells and targeted static properties and nonlinear responses, e.g. [3, 4]. The design of multi-unit-cell structures under dynamic loading conditions has thus far received only limited attention. This design of mechanical metamaterials for high-rate applications necessitates a framework that is sufficiently flexible to accommodate a combination of diverse unit cells. Achieving this flexibility necessitates a combination of different unit cells with a common architecture. This is enabled by breaking down the fundamental properties of each unit cell into scalar features. A map of these features can then be used as a representation of a larger structure containing unit cells connected without discontinuities. Generative design methodologies, including diffusion models, present a promising approach to generating these maps with the desired response. Diffusion models of this nature must be conditioned on the stress-strain rate behavior at varying strain rates to facilitate the targeted design of structures for a range of applications. This investigation demonstrates the capacity of diffusion models to generate targeted stress-strain curves for a composition of unit cells into a larger structure. The calibration of the inverse model is enabled by the employment of structural finite elements along with a machine learning framework informed by geometric aspects of the structure. The integration of these physical and data-driven methodologies facilitates the development of lightweight, heterogeneous structures that exhibit the desired responses. [1] https://doi.org/10.1007/978-3-031-93213-7_1 [2] https://doi.org/10.1038/s43588-024-00672-x [3] https://doi.org/10.48550/arXiv.2507.15753 [4] https://doi.org/10.1016/j.cma.2025.118499