A multi-modal AI demonstration system based on STAR model, supporting text-to-image generation, image editing, and image understanding.
Controls how closely the generated image matches the text
Number of tokens to consider during sampling
Nucleus sampling parameter