Large Language Model FundamentalsCornell

Certificate Program

In this course, you will discover how to work directly with some of today's most powerful large language models (LLMs). You'll start by exploring online LLM-based systems and seeing how they handle tasks ranging from creative text generation to language translation. You'll compare how models from major organizations like OpenAI, Google, and Anthropic differ in their outputs and underlying philosophies.

You will then move beyond web interfaces to identify how to find and load various foundation models through the Hugging Face hub. By mastering Python scripts that retrieve and run these models locally, you'll gain deeper control over prompt engineering and understand how different model architectures respond to your requests. Finally, you'll tie all these skills together in hands-on projects where you generate text, analyze tokenization details, and assess outputs from multiple LLMs.

In this course, you will use Python to quantify the next-word predictions of large language models (LLMs) and understand how these models assign probabilities to text. You'll compare raw scores from LLMs, transform them into probabilities, and explore uncertainty measures like entropy. You'll also build n-gram language models from scratch, apply smoothing to handle unseen words, and interpret log probabilities to avoid numerical underflow.

By the end of this course, you will be able to evaluate entire sentences for their likelihood, implement your own model confidence checks, and decide when and how to suggest completions for real-world text applications.

You are required to have completed the following course or have equivalent experience before taking this course:

• LLM Tools, Platforms, and Prompts

In this course, you will discover how to adapt and refine large language models (LLMs) for tasks beyond their default capabilities by creating curated training sets, tweaking model parameters, and exploring cutting-edge approaches such as preference learning and low-rank adaptation (LoRA). You'll start by fine-tuning a base model using the Hugging Face API and analyzing common optimization strategies, including learning rate selection and gradient-based methods like AdaGrad and ADAM.

As you progress, you will evaluate your models with metrics that highlight accuracy, precision, and recall, then you'll extend your techniques to include pairwise preference optimization, which lets you incorporate direct user feedback into model improvements. Along the way, you'll see how instruction-tuned chatbots are built, practice customizing LLM outputs for specific tasks, and examine how to set up robust evaluation loops to measure success.

By the end of this course, you'll have a clear blueprint for building and honing specialized models that can handle diverse real-world applications.

You are required to have completed the following courses or have equivalent experience before taking this course:

• LLM Tools, Platforms, and Prompts
• Language Models and Next-Word Pronunciation

In this course, you will analyze how large language models are constructed from diverse text sources and examine the entire model life cycle, from pretraining data collection to generating meaningful outputs. You'll explore how choices about data type, genre, and tokenization affect a model's performance, discovering how to compare real-world corpora such as Wikipedia, Reddit, and GitHub.

Through hands-on projects, you will design tokenizers, quantify text characteristics, and apply methods like byte-pair encoding to see how different preprocessing strategies shape model capabilities. You'll also investigate how models interpret context by studying keywords in context (KWIC) views and embedding-based analysis.

By the end of this course, you will have a clear understanding of how data selection and processing decisions influence the way LLMs behave, preparing you to evaluate or improve existing models.

You are required to have completed the following courses or have equivalent experience before taking this course:

• LLM Tools, Platforms, and Prompts
• Language Models and Next-Word Pronunciation
• Fine-Tuning LLMs

In this course, you will investigate the internal workings of transformer-based language models by exploring how embeddings, attention, and model architecture shape textual outputs. You'll begin by building a neural search engine that retrieves documents through vector similarity then move on to extracting token-level representations and visualizing attention patterns across different layers and heads.

As you progress, you will analyze how tokens interact with each other in a large language model (LLM), compare encoder-based architecture with decoder-based architectures, and trace how a single word's meaning can shift from input to output. By mastering techniques like plotting similarity matrices and identifying key influencers in the attention process, you'll gain insights enabling you to decode model behaviors and apply advanced strategies for more accurate, context-aware text generation.

You are required to have completed the following courses or have equivalent experience before taking this course:

• LLM Tools, Platforms, and Prompts
• Language Models and Next-Word Pronunciation
• Fine-Tuning LLMs
• Language Models and Language Data