How can the Effectiveness of Vision Transformers be Leveraged in Diffusion-based Generative Learning? This Paper from NVIDIA Introduces a Novel Artificial Intelligence Model Called Diffusion Vision Transformers (DiffiT)