Deep Learning Developments with PyTorch

Catalog description

PyTorch stands out as a leading machine learning framework built upon the Torch library. Renowned for its adaptability and user -friendly interface, PyTorch has garnered a substantial following in both industry and academia, becoming the go -to choice for ma ny cutting -edge research projects. Its popularity is evidenced by the fact that a significant portion of modern research code is written using PyTorch. In this course, we offer a comprehensive exploration of PyTorch, providing a step -by-step guide to its m odern applications. The course content is structured to cover a wide array of topics, primarily focusing on three prominent application areas: computer vision, natural language processing, and reinforcement learning. Throughout the course, students will de lve into the intricacies of utilizing PyTorch for various tasks, including but not limited to image/video classification, object detection, semantic segmentation, text classification, sequence-to-sequence translation, visual question answering, and Deep Q -Networks (DQN) for reinforcement learning. Moreover, the course delves into modern deep learning architectures, providing insights into 2D/3D convolutional neural networks (CNNs), recurrent neural networks (RNNs), long short - term memory (LSTM) networks, transformers, and encoder-decoder networks. By thoroughly exploring these architectures, students will gain a solid understanding of how to leverage PyTorch effectively for state-of-the-art deep learning applications across diverse domains. MAI660 Large Language Models: Theory and Practice (3 Credit Hours) A recent advancement in neural network technology, known as the large language model (LLM), has been making waves in the media lately. Notably, ChatGPT and Microsoft's new Bing Chat interface have been grabbing headlines regularly. This course serves as a comprehensive introduction to this cutting -edge technology, delving into its historical roots in computational linguistics and language modeling and examining the foundational design principles that drive its implementation in modern AI systems. Throughout the course, students will delve into various aspects of LLM technology, gaining insights into language modeling, the attention mechanism, prompt and instruction tuning, composability, quantization, low-rank adaptation, and a plethora of software and hardw are optimizations. These optimizations are crucial for enabling LLMs to operate at scale while maintaining acceptable latencies, making them viable for real -world applications in diverse domains. By the end of the course, students will have a deep understanding of the intricacies involved in leveraging LLMs effectively in AI systems. MAI 680A Game Design and Analysis (3 Credit Hours) Provides theoretical background and foundation for analyzing and designing games. Examines fundamental domains that are necessary to understand what games are and how they affect players, including but not limited to interface design, level design, narrative, learning, and culture. Presents relevant concepts and frameworks from a wide variety of disciplines —psychology, phenomenology, sociology, anthropology, media studies, affect theories, learning theories, and theories of motivation—for each domain. Explains the core elements of game design, introduces students to formal abstract design tools, explores several models of design process and iteration, and offers students an opportunity to practice game design in groups. MAI 680B AI Generative (3 Credit Hours) The AI Generative course immerses learners in the essential knowledge and practical skills needed to fully harness the potential of this groundbreaking technology. It is meticulously designed to equip participants with comprehensive expertise in artificial intelligence, particularly focusing on Generative AI —a domain where machines demonstrate the remarkable ability to create art and other creative outputs.