Course Materials

# Module 1

Getting Started: Google Colab and GitHub

Learning objectives: Open, run, and save a notebook in Google Colab; navigate a GitHub repository and open a notebook from it; describe the course workflow for accessing, editing, and retaining copies of code.

Before class

• Sign in to Google ColabOptional

During class

• Sign in to or create a GitHub accountRequired
• Practice opening, running, copying, and saving a course notebook.Required

After class

• Code.org, “How Chatbots and Large Language Models Work” (about 7 min)Required
• Karpathy, “Intro to Large Language Models” (1 hr, listenable as a podcast)Optional

# Module 2

Tokens and Embeddings

Learning objectives: Tokenize short texts and interpret token IDs; use sentence embeddings to compare semantic similarity; distinguish representation models from text-generation models; explain how next-token probabilities produce generated text.

Before class

• 3Blue1Brown, “But what is a GPT? Visual intro to transformers” (27 min)Required
• Read Chapter 1, “An Introduction to Large Language Models,” and Chapter 2, “Tokens and Embeddings.”Optional

During class

• Reference notebook: Introduction to Language ModelsOptional
• Reference notebook: Tokens and Token EmbeddingsOptional

After class

• Hugging Face LLM Course, “Tokenizers”Optional
• Jay Alammar, “The Illustrated Word2vec”Optional
• TensorFlow Embedding Projector: explore nearest neighbors and semantic clusters in a pretrained embedding spaceOptional

# Module 3

Text Classification I: Encoder Models

Learning objectives: Distinguish binary, multiclass, and multilabel classification problems; apply task-specific and embedding-based classifiers; evaluate predictions using suitable metrics and structured error analysis.

Before class

• Kim, Soderman, and Sang, “Sentiment analysis of solar energy in U.S. Cities: a 10-year analysis using transformer-based deep learning”: read sections 1, “Introduction,” and 2, “Materials and approach,” onlyRequired
• Jay Alammar, “The Illustrated BERT”: read “Example: Sentence Classification” through “Model Outputs”Optional
• IBM Technology, “Text Classification: AI Techniques and Real-World Applications” (14 min)Required
• StatQuest, “Machine Learning Fundamentals: The Confusion Matrix” (7 min)Required
• 3Blue1Brown, “Attention in transformers, step-by-step”Optional
• Read Chapter 3, “Looking Inside Large Language Models,” and the Chapter 4 sections on task-specific and embedding-based text classification.Optional

During class

• Reference notebook: Looking Inside LLMsOptional
• Companion notebook: Text ClassificationOptional

After class

• Hugging Face LLM Course, “How Transformers Solve Tasks”: read “Text classification”Optional
• Hugging Face LLM Course, “Working with pipelines”: try the sentiment-classification example with your own sentencesOptional
• Hugging Face Transformers, “Text classification”: fine-tune DistilBERT on IMDb and run inferenceOptional
• Hugging Face SetFit, “Sentence Transformers Finetuning”: see how an embedding model and classification head work togetherOptional
• TensorFlow Playground: run a classification problem and observe how the learned decision boundary changesOptional

# Module 4

Text Classification II: Zero-Shot and Few-Shot Models

Learning objectives: Compare NLI-based zero-shot classification with prompted generative classification; use label descriptions, few-shot examples, and constrained outputs; test how label wording and examples affect classification performance.

Before class

• Read Chapter 4, “Text Classification,” from “What If We Do Not Have Labeled Data?” through “Text Classification with Generative Models,” including the Flan-T5 and ChatGPT classification examples.Optional
• IBM Technology, “What is Zero-Shot Learning?” (9 min)Required
• CodeEmporium, “Few Shot Learning – explained!” (10 min)Required

During class

• Companion notebook: Text ClassificationRequired

After class

• Bhavesh Bhatt, “Zero Shot Classification with HuggingFace Pipeline” (hands-on code walkthrough)Optional
• Hugging Face, “Zero-Shot Text Classification”Optional

# Module 5

Text Clustering and Topic Modeling

Learning objectives: Create document embeddings for an unlabeled text collection; cluster and visualize semantically related documents; interpret topic representations while identifying instability, labeling choices, and other limitations.

Before class

• Chapter 5: “Text Clustering and Topic Modeling”Optional

During class

• Companion notebook: Text Clustering and Topic ModelingOptional

# Module 6

Context Design, Model Benchmarks, and Evaluation

Learning objectives: Construct prompts with clear instructions, examples, context, and output constraints; distinguish model benchmarks from application and system evaluations; compare deterministic checks, rubrics, pairwise judgments, and human review; explain limitations involving prompt sensitivity, contamination, saturation, judge bias, latency, and cost.

Before class

• Google Cloud Tech, “Prompt Engineering for Developers” (6 min; includes zero-shot, one-shot, and few-shot prompting)Required
• Google AI for Developers, “Prompt Design Strategies” (zero-shot and few-shot sections)Optional
• Stanford HELM: Holistic Evaluation of Language ModelsRequired
• Chapter 6: “Prompt Engineering”Optional

During class

• Mini evaluation lab: define grading criteria, compare two model or prompt conditions, and inspect disagreements among deterministic checks, rubric judgments, and human reviewRequired
• Companion notebook: Prompt EngineeringOptional

# Module 7

RAG I: Dense Retrieval and Grounded Generation

Learning objectives: Chunk and embed a small document collection; build a dense semantic-retrieval workflow; use retrieved passages to produce grounded answers with source citations; distinguish retrieval failures from unsupported generation.

Before class

• Microsoft Learn, “Introduction to Retrieval-Augmented Generation Concepts”Required
• Selected sections from Chapter 8: dense retrieval and retrieval-augmented generationOptional

During class

• Guided lab: chunk and embed a document collection, retrieve relevant passages, and generate cited answers while tracing retrieval and generation failuresRequired
• Companion notebook: Semantic Search and Retrieval-Augmented GenerationOptional

# Module 8

RAG II: Hybrid Retrieval, Reranking, and Evaluation

Learning objectives: Compare lexical, dense, and hybrid retrieval; use a cross-encoder to rerank candidate passages; evaluate retrieval and answer quality; explain tradeoffs involving accuracy, latency, complexity, and cost.

Before class

• Sentence Transformers, “Retrieve & Re-Rank”Required
• Microsoft, “Information Retrieval for a RAG Solution”Optional

During class

• Guided lab: compare dense retrieval with a lexical or hybrid baseline, rerank candidates, and measure retrieval and grounded-answer qualityRequired

# Module 9

LLM Workflows, Tool Use, and Agentic RAG

Learning objectives: Distinguish deterministic workflows from model-directed agents; trace the tool-call, execution, observation, and response loop; explain how context, permissions, stopping conditions, and retrieval-as-a-tool affect reliability; identify failures from a supplied execution trace.

Before class

• Microsoft Developer, “AI Agents for Beginners” — selected segments: 0:00–5:55 and 19:35–38:54 (about 25 min)Required
• Hugging Face Agents Course, “What Are Tools?”Optional
• Chapter 7: “Advanced Text Generation Techniques and Tools”Optional

During class

• Online activity: analyze successful and failed tool-use traces, identify the first failure, check the actual task outcome, and propose a safeguardRequired
• Companion notebook: Advanced Text Generation Techniques and ToolsOptional

# Module 10

Model Adaptation I: Text Embedding Models

Learning objectives: Explain when a general embedding model needs task or domain adaptation; organize query-positive pairs and hard negatives for contrastive learning; fine-tune a small embedding model; compare the original and adapted models on held-out retrieval examples.

Before class

• Sentence Transformers, “Training Overview”Required
• Chapter 10: “Creating Text Embedding Models”Optional

During class

• Guided lab: fine-tune a small embedding model with query-positive pairs and hard negatives, then compare before-and-after retrieval on held-out queriesRequired
• Companion notebook: Creating Text Embedding ModelsOptional

# Module 11

Model Adaptation II: Parameter-Efficient Fine-Tuning

Learning objectives: Choose among prompting, retrieval, embedding adaptation, and generative fine-tuning; prepare instruction or conversational data with the correct chat template; explain LoRA and QLoRA; compare a base and adapted model while assessing overfitting, privacy, computing, and maintenance risks.

Before class

• Hugging Face PEFT, “LoRA”Required
• Hugging Face Transformers, “Chat Templates”Optional
• Selected sections from Chapter 12, “Fine-Tuning Generation Models”: supervised fine-tuning, LoRA/QLoRA, instruction data, and evaluationOptional

During class

• Instructor overview: deciding among prompting, retrieval, embedding adaptation, and generative fine-tuningRequired
• Guided lab: fine-tune a small model with LoRA or QLoRA and compare the base and adapted models on held-out task examplesRequired
• Reference: Hugging Face TRL SFT TrainerOptional
• Reference notebook: Fine-Tuning Generation ModelsOptional

# Module 12

Project Presentations

Learning objectives: Present the project question, data, workflow, evaluation, and results as a coherent analytical story; use clear visuals or examples to explain model behavior; discuss limitations and responsible use and respond thoughtfully to questions and peer feedback.

Before class

• Review the project presentation requirementsRequired
• Confirm that your GitHub repository opens correctly, the README identifies the starting point, and the analysis can be demonstrated during questions.Required
• Submit the required presentation materials through the designated Canvas link before the presentation window closes.Required