CSD: Large Language Models

This course will focus on understanding the potential opportunities, challenges, and guardrails of LLMs. Participants will first gain an understanding of the foundations of these models, work on creating their own models and work with new tools that can help design robust and resilient AI models. Participants will be introduced to diverse domains where language models play a part. Student will be introduced to robust and resilient development tools along with techniques to reduce the environmental impact of LLMs. Participants will also learn about the moral and ethical issues around LLMs and be presented with techniques to interrogate new systems and mitigate such issues with careful system design and planning. Finally, the various efforts underway to develop guardrails for LLMs will be explored. This project aims to address the key questions associated with understanding advanced mobile communications networks from societal, technological, and security perspectives. The associated workshop will provide a venue validating and obtaining feedback on these ideas. The questions include (but are not limited to):

1. What types of methods are being developed to support large language models? For example, are there particular mathematical concepts that are common across societal, technological, and security domains?
2. Are there opportunities to improve the performance of these methods kernels or algorithms using cyberinfrastructure and software tools? If so, how should these tools be shared with domain practitioners?
3. What are future trends of large language models and what are their societal, technological, and security implications?
4. How do practitioners who wish to use large language models for analysis and discovery adapt existing trained large language models to their own datasets and problems?
5. What forms of cyberinfrastructure (e.g., software tools) would make practitioners most efficient in applying large language models to different domains?
6. What opportunities exist to combine state-of-the-art techniques from optimization, machine learning, network science and high-performance computing to further develop large language models?

This course is by invitation only. Invitations are issued based on collaborations and not on an individual basis. If you are currently enrolled, you can view the course content by visiting your Dashboard. To access your Dashboard, click on your username on the toolbar at the top of your screen and select it from the dropdown menu.