The field of additive manufacturing has seen significant advancements, which in turn drove the development of high-performance cellular solids, including truss-, shell-, and plate-lattices. Among these, plate-lattices exhibit promising mechanical properties, exhibiting optimal mass-specific stiffness [1]. Plate-lattices are broadly defined by the arrangement of their constituent plates with translational periodicity. This research presents a method for constructing general plate-lattices and a neural network-based approach to solve the inverse design problem—mapping desired properties to lattice structures. The plate-lattices are created by the sequentially positioning of plates within a unit cell by selecting an anchor point, a normal vector and a thickness, all adhering to periodic boundary conditions [2]. We construct a database of over 90’000 plate-lattices and analyse their effective anisotropic stiffness tensor, using finite element simulations. We show that the flexible arrangement of plates allows for geometric control over a wide range of macroscopic properties, while using the same amount base material. Employing this general framework, we describe a correlation between plate orientation and stiffness and provide a comprehensive comparison to the property spaces of truss- and shell-lattices [3]. The inverse model uses LSTM and feedforward layers, with carefully chosen activation functions to ensure physically meaningful outputs. The forward model parameters are optimized through supervised training, while the inverse model is trained alongside a pre-trained structure-to-property model. Additionally, we perform a comprehensive grid search to optimize hyperparameters, including the number of layers and hidden state sizes. We evaluate the model's ability to generalize by testing its performance on unseen data. The model demonstrated strong performance, allowing for the efficient customization of the stiffest category of metamaterials. [1] Tancogne‐Dejean, Thomas, et al. "3D plate‐lattices: an emerging class of low‐density metamaterial exhibiting optimal isotropic stiffness." Advanced Materials 30.45 (2018): 1803334. [2] Meyer, Paul P., Thomas Tancogne-Dejean, and Dirk Mohr. "Non-symmetric plate-lattices: Recurrent neural network-based design of optimal metamaterials." Acta Materialia 278 (2024): 120246. [3] Meyer, Paul P., et al. "Graph-based metamaterials: Deep learning of structure-property relations." Materials & Design 223 (2022): 111175.