Text-to-image and image-to-text with one model
Generate images from text prompts
Generate images from text descriptions