Precise histologic pattern classification is essential for lung adenocarcinoma, as it is a primary cause of cancer-related death, to inform successful treatment approaches. The objective of this paper is to compare several deep neural network designs for the categorization of lung cancer histologic patterns. We test various models like DeiT (Data Efficient Image Transformers), CAiT, Swin Transformer, ViT, ResNet, employing Cohen Kappa Score and accuracy measures using hematoxylin and eosin (H&E)-stained formalin-fixed paraffin-embedded (FFPE) whole-slide images of lung adenocarcinoma from the Dartmouth-Hitchcock Medical Center (DHMC) [17]. Our findings highlight the results of each architecture, offering guidance on whether models are suitable for tasks involving the classification of histopathology images.