Microsoft Lens 3.8B-parameter text-to-image diffusion model
SMRTR summary
Microsoft's Lens is a 3.8 billion-parameter AI model that generates high-resolution images from text descriptions. Trained on 800 million image-text pairs with detailed captions, it produces competitive image quality using less computing power than larger models, and supports multiple aspect ratios up to 1440×1440 pixels.
SMRTR provides this summary for quick context. The original article belongs to Hacker News.
Read the original article