Large Language Models Methods and Application

12 units · Graduate

Fall: 12 units This course provides a broad foundation for understanding, working with, and adapting existing tools and technologies in the area of Large Language Models like BERT, T5, GPT, and others. It begins with a short history of the area of language models and quickly transitions to a broad survey of the area, offering exposure to the gamut of topics including systems, data, data filtering, training objectives, RLHF/instruction tuning, ethics, policy, evaluation, and other human facing issues. Students will delve into Transformer architectures more broadly and how they work, as well as exploring the reasons why they are better than LSTM-based seq2seq, decoding strategies, etc. Students will learn through readings and hands-on assignments where they will explore techniques for pretraining, attention, prompting, etc. They will then apply these skills in a semester-long course project, making use of locally sourced model instances that offer the opportunity to explore behind the curtain of commercial APIs. Prerequisites: 11-711 or 11-685 or 11-785 or 10-601 or 10-701 Course Website: https://cmu-llms.org/syllabus/