Skip to main content

Ctrl+K

Course overview: Understanding LLMs

01 Introduction

Background
Sheet 1.1: Practical set-up & Training data

02 ANNs & RNNs

PyTorch, ANNs & LMs
Sheet 2.1: PyTorch essentials
Sheet 2.2: ML-estimation
Sheet 2.3: Non-linear regression (MLP w/ PyTorch modules)
Sheet 2.4: Character-level sequence modeling w/ RNNs
Sheet 2.5: Introduction to HuggingFace & LMs

03 LSTMs & transformers

LSTMs & Transformers
Sheet 3.1: Tokenization & Transformers
Sheet 3.2: Transformer configurations & Training utilities

04 Prompting

Prompting & Current LMs
Sheet 3.3: Prompting & Decoding

05 Fine-tuning & RLHF

Fine-tuning and RLHF
Sheet 4.1 Supervised fine-tuning and RL fine-tuning

06 Agents & applications

LLM systems & agents
Sheet 5.1 LLM agents

07 Attribution

Attribution methods
Sheet 6.1 LLM probing & attribution

08 Behavioral evaluation

Evaluation & behavioral assessment
Sheet 7.1: Behavioral assessment & Evaluation
Sheet 7.2: Advanced evaluation

09 Mechanistic interpretation

Mechanistic Interpretability
Sheet 8.1: Mechanistic interpretability

10 Understanding LLMs

Implications, Understanding & Philosophy

Homework

Homework 1: Language models (50 points)
Homework 2: Prompting & Generation with LMs (50 points)
Homework 3: LLM agents & RL fine-tuning
Homework 4: LLM evaluation

Repository
Open issue

Index

By Michael Franke, Carsten Eickhoff, Polina Tsvilodub

© Copyright 2024.