CS 839 Fall 2024: 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

x -->

Date	Lecture	Readings	Homework Released	Homework Due
Thursday Sept. 5	Introduction and Course Overview
Tuesday Sept. 10	Machine Learning Mini-Review	Patterns, Predictions, and Actions
Thursday Sept. 12	Architectures I: Transformers & Attention	Attention Is All You Need The Illustrated Transformer
Tuesday Sept. 17	Architectures II: Subquadratic Architectures	Mamba: Linear-Time Sequence Modeling with Selective State Spaces Mamba: The Hard Way
Thursday Sept. 19	Models I: Encoder-Decoder, Encoder-Only	BERT Paper RoBERTa Paper T5 Paper	HW 1 Released
Tuesday Sept. 24	Models II: Decoder-Only. Start Prompting I: Basics	GPT-3 Paper Llama 3.1 Paper
Thursday Sept. 26	Prompting I: Basics, In-Context Learning, Improvements, Extensions	The Prompt Report: A Systematic Survey of Prompting Techniques Finetuned Language Models Are Zero-Shot Learners
Tuesday Oct. 1	Prompting II: Chain-of-Thought, Advanced Techniques	CoT Paper Large Language Models are Zero-Shot Reasoners Tree of Thoughts
Thursday Oct. 3	Specialization: Fine-Tuning, Adaptation, Editing	Low-Rank Adaptation of Large Language Models Fast Model Editing at Scale		HW 1 Due
Tuesday Oct. 8	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	Presentation Info Released
Thursday Oct. 10	Efficient Training	FlashAttention Megatron-LM	HW 2 Released
Tuesday Oct. 15	Efficient Inference	Fast Inference from Transformers via Speculative Decoding Efficient Memory Management for Large Language Model Serving with PagedAttention
Thursday Oct. 17	Guest Lecture On Hybrid Foundation Models
Tuesday Oct. 22	Data	The Pile Deduplication Helps LMs	Project Information Out
Thursday Oct. 24	Benchmarks and LLM-as-Judge	Judging LLM-as-a-Judge with MT-Bench and Chatbot Arena Can Large Language Models Really Improve by Self-critiquing Their Own Plans?
Tuesday Oct. 29	Multimodal Foundation Models	CodeGen Paper Flamingo Paper AlphaFold Paper ClimaX Paper LegalBench Paper	HW 3 Released
Thursday Oct. 31	Scaling & Scaling Laws	"Compute-Optimal" LLMs Scaling Laws for Neural LMs		HW2 Due (Extended)
Tuesday Nov. 5	Diffusion Models	Denoising Diffusion Probabilistic Models Maximum Likelihood Training of Score-Based Diffusion Models
Thursday Nov. 7	Security, Privacy, Toxicity + Future Areas	Extracting Training Data from LLMs LLMs Can Be Strong Differentially Private Learners Red Teaming Paper
To Dec. 5	Student Presentations			Due: HW 3, 1/2 page info on presentations