This model is trained on a sample of CheXpert Dataset using only frontal views. The model combines a ViT and GPT2 to generate draft radiology reports.