Foundations of Generative AI

Catalog description

Design and implementation of generative AI systems, covering models for generating text, images, audio, and multimodal data. Includes transformer-based approaches (e.g., large language models) as well as non-transformer methods and generative adversarial networks (GANs), featuring generator-discriminator training dynamics, and variational autoencoders (VAEs) based on probabilistic latent-variable modeling and reconstruction. Topics include self-attention, self-supervised learning, model training and evaluation, fine-tuning, and key challenges such as hallucinations, bias, scalability, and responsible AI use, with hands-on implementation.