Hacker News·2yDIAMOND
DIAMOND 💎 (DIffusion As a Model Of eNvironment Dreams) is a reinforcement learning agent trained using a diffusion world model, showing improved performance by preserving important visual details. Highlighting its use in 3D environments like CSGO, DIAMOND achieves a mean human-normalized score of 1.46 on the Atari 100k benchmark. By using an efficient diffusion model, DIAMOND sets a new standard for agents trained entirely within world models.
Towards AI·2y[AI/ML] Diffusion Models — A Beginner’s Guide to Math Behind Stable Diffusion and Dall-e!
Diffusion Models are revolutionizing generative modeling in computer vision, especially through tools like DALL-E and Stable Diffusion. These models add and remove noise to and from images across multiple steps, enhancing image generation quality. Key mathematical perspectives include Markov Chains and Langevin Dynamics. The architecture commonly involves U-Net and various conditioning methods, such as classifier-guided and classifier-free guidance. Enhancements to these models, like the use of ControlNet and improvements in sampling techniques, make them more efficient and versatile for generating high-quality images.
Towards Data Science·2yThe Evolution of Text to Video Models
Text-to-video generation is significantly more complex than text-to-image, demanding understanding of object movement and temporal consistency. Modern video diffusion models, like VDM, Make-A-Video by Meta AI, Imagen Video, and SORA, tackle these challenges using strategies such as combining image-text and unlabelled video data, spatial and temporal layers, and latent diffusion. Large-scale datasets and computational advancements are expected to drive future innovations in this field.