A textAvatarsimage Transformer model that achieves state-of-the-art image generation performance while being significantly more efficient than diffusion or autoregressive models.Muse: Text-To-Image Generation via Masked Generative Transformers