CS 839 Fall 2025: Foundation Models

Important Notes

This schedule is tentative and subject to change. Please check back often. In particular, the deadlines for the homework sets/project may change; please see Canvas for the actual deadlines.

Tentative Schedule

-->

Date	Lecture	Readings	Homework Release	Homework Due
Thursday Sept. 4	Introduction and Course Overview
Tuesday Sept. 9	Machine Learning Mini-Review	Patterns, Predictions, and Actions
Thursday Sept. 11	Architectures I: Transformers & Attention	Attention Is All You Need The Illustrated Transformer Tokenization RoFormer: Rotary Position Embedding (RoPE) ALiBi: Train Short, Test Long
Tuesday Sept. 16	Architectures II: Subquadratic Architectures	Mamba: Linear-Time Sequence Modeling with Selective State Spaces Mamba: The Hard Way Retentive Network (RetNet) Transformers are SSMs
Thursday Sept. 18	Models I: Encoder-Decoder, Encoder-Only	BERT Paper RoBERTa Paper T5 Paper	HW 1 Released
Tuesday Sept. 23	Models II: Decoder-Only. Start Prompting	GPT-3 Paper Llama 3.1 Paper
Thursday Sept. 25	Prompting: Basics, In-Context Learning, Chain-of-Thought	Finetuned Language Models Are Zero-Shot Learners Learning without training: The implicit dynamics of in-context learning CoT Large Language Models are Zero-Shot Reasoners
Tuesday Sept. 30	Specialization: Fine-Tuning, Adaptation, Editing	Low-Rank Adaptation of Large Language Models Fast Model Editing at Scale QLoRA: Efficient Finetuning of Quantized LLMs
Thursday Oct. 2	Alignment: RHLF, DPO, and friends	Hugging Face RLHF Blog Training language models to follow instructions with human feedback Direct Preference Optimization: Your Language Model is Secretly a Reward Model Constitutional AI: Harmlessness from AI Feedback RLAIF: Scaling RLHF with AI Feedback	Presentation Info Released	HW 1 Due
Tuesday Oct. 7	Reinforcement Learning with Verifiable Rewards (RLVR)	RLHF Book (Ch. 14, Ch. 11.1.4)	HW 2 Released
Thursday Oct. 9	Efficient Training	FlashAttention Megatron-LM
Tuesday Oct. 14	Efficient Inference	Fast Inference from Transformers via Speculative Decoding Efficient Memory Management for Large Language Model Serving with PagedAttention LLM.int8() AWQ Quantization Medusa Decoding
Thursday Oct. 16	Evaluation, Benchmarks, and LLM-as-a-Judge	HELM: Holistic Evaluation of Language Models MMLU-Pro LLM-as-a-Judge: MT-Bench & Chatbot Arena
Tuesday Oct. 21	Agents	ReAct: Reasoning + Acting Toolformer Reflexion: Verbal RL for Agents AgentBench: Evaluating LLMs as Agents OSWorld: Multimodal Computer Use Benchmark WebArena: Realistic Web Environment Voyager: Embodied Agent in Minecraft		HW2 Due
Thursday Oct. 23	Reasoning, Verifiers, Rubrics	Self-Consistency Improves Chain-of-Thought Tree of Thoughts Let’s Verify Step by Step (Process Supervision) Program-of-Thoughts Prompting OpenAI o1 System Card (Reasoning models) DeepSeek-R1: Incentivizing Reasoning via RL		Presentation proposal due
Tuesday Oct. 28	Multimodal Foundation Models	CLIP: Learning Transferable Visual Models Flamingo LLaVA: Visual Instruction Tuning SAM 2: Segment Anything in Images & Videos MMMU: Multimodal Understanding & Reasoning AlphaFold ClimaX	Project info released
Thursday Oct. 30	Diffusion Models	Denoising Diffusion Probabilistic Models Maximum Likelihood Training of Score-Based Diffusion Models Diffusion language models Score-based Generative Modeling via SDEs Flow Matching for Generative Modeling Consistency Models
Tuesday Nov. 4	Scaling & Scaling Laws	"Compute-Optimal" LLMs Scaling Laws for Neural LMs
Thursday Nov. 6	Security, Privacy, Toxicity + Future Areas	Extracting Training Data from LLMs LLMs Can Be Strong Differentially Private Learners Red Teaming Paper LLM-based agentic systems
To Dec. 9 (Note: no class Dec. 4)	Student Presentations			Due: HW 3, 1/2 page info on presentations