llm-course

Free

Course to get into Large Language Models (LLMs) with roadmaps and Colab notebooks.

Model APIsFreeFree tier

Type

Open Source

Visit Website GitHub (79.0k stars)

About llm-course

LLM Course is a collection of in-depth blog articles and tutorials by Maxime Labonne covering the entire spectrum of Large Language Models (LLMs), from beginner concepts to advanced techniques. The content includes hands-on Colab notebooks for fine-tuning Llama models, quantization methods (GPTQ, GGUF, EXL2), model merging with MergeKit, decoding strategies, and more. The blog also features chapters on graph neural networks and optimization topics, providing a comprehensive resource for anyone looking to gain practical expertise in LLMs.

Key Features

Covers fine-tuning of Llama 2, Llama 3, and Mistral-7B with techniques like ORPO and DPO

Guides on quantization using GPTQ, GGUF/llama.cpp, and ExLlamaV2

Tutorials on model merging with MergeKit and creating Mixture of Experts (MoE)

Explains decoding strategies including beam search, nucleus sampling, and more

Includes practical Colab notebooks for hands-on learning

Topics range from beginner to advanced including agentic data generation and abliteration

Also covers graph neural networks (GCN, GraphSAGE, GAT) and mathematical optimization

Pros & Cons

Pros

Completely free and open-access blog with no paywalls
Hands-on Colab notebooks for each major technique
Covers state-of-the-art methods (e.g., ORPO, abliteration, AgentInstruct)
Written by a known expert in the LLM community (Maxime Labonne)
Broad scope from basics to frontier research

Cons

Not a structured curriculum with assignments or assessments
Relies on blog posts; no interactive community or support forum
Some topics assume prior knowledge of machine learning and Python
Content may become outdated as LLM field rapidly evolves

Best For

Self-paced learning for AI practitioners and researchers wanting to master LLMsPractical reference for fine-tuning and quantizing open-source LLMsUnderstanding model merging and MoE architecturesExploring advanced LLM topics like agentic data generation and reinforcement learningSupplementing formal courses on natural language processing and deep learning

Alternatives to llm-course

AutoGPT

AutoGPT is the vision of accessible AI for everyone, to use and to build on. Our mission is to provide the tools, so that you can focus on what matters.

aider

aider is AI pair programming in your terminal

axolotl

Go ahead and axolotl questions

Chroma

Open-source embedding database

awesome-claude-code

A curated list of awesome skills, hooks, slash-commands, agent orchestrators, applications, and plugins for Claude Code by Anthropic

agents-course

This repository contains the Hugging Face Agents Course.

FAQ

Does this course include hands-on notebooks?

Yes, many tutorials provide Colab notebooks, such as 'Fine-Tune Your Own Llama 2 Model in a Colab Notebook' and 'Fine-tune Llama 3.1 Ultra-Efficiently with Unsloth'.

What LLM fine-tuning techniques are covered?

The blog covers supervised fine-tuning with tools like Unsloth, ORPO (Odds Ratio Preference Optimization), Direct Preference Optimization (DPO), and abliteration.

Are there tutorials on model quantization?

Yes, the blog explains GPTQ, GGUF/llama.cpp, and ExLlamaV2 quantization, including 4-bit and 8-bit methods.

Does the course cover model merging?

Yes, it provides a guide on merging LLMs with MergeKit and creating Mixture of Experts (MoE) models.