Meta has revealed its new AI-powered art generator, CM3Leon or 'Chameleon', which claims to achieve state-of-the-art performance for text-to-image generation. CM3Leon is unique from other AI image generators in creating captions for images, which will lay the groundwork for more capable image-understanding models. In a blog post shared with Tech Crunch, Meta wrote, "With CM3Leon’s capabilities, image generation tools can produce more coherent imagery that better follows the input prompts. We believe CM3Leon’s strong performance across a variety of tasks is a step toward higher-fidelity image generation and understanding.” Most AI image generators like DALL-E2 and Imagen, use diffusion to generate images by subtracting noise from the starting image, which is quite an expensive tech. Meta's image generator uses a mechanism called “attention” to weigh the relevance of input data such as text or images. This boosts model training speed and makes them easily parallelizable. Meta claims that CM3leon only requires five times less computing and a smaller training dataset than previous transformer-based methods. The company used millions of data points from Shutterstock to train the AI model. CM3leon is more advanced than other AI image models generating images from complex prompts with ease. The AI program can edit existing images with instructions and performs better than DALL-E2. Meta hasn't released a statement informing of its program release yet.
from Latest Technology News, Tech News Pakistan | The Express Tribune https://ift.tt/5DVcwIo
via IFTTT
0 Comments