If you want to understand the underlying principles of large language models, most resources only introduce theoretical knowledge or provide source code, leaving you confused.

I stumbled upon the open-source tutorial EveryonesLLM, which guides you step by step to build a complete large language model from scratch on Google Colab, writing all the code yourself.

The entire tutorial is divided into 29 chapters, starting from basic data loading and word embeddings, gradually building up to attention mechanisms, Transformer modules, and finally completing pretraining and instruction fine-tuning.

GitHub: http://github.com/HayatoHongo/EveryonesLLM

Each chapter is an independent Colab notebook — just open it in your browser and run it; no need to fuss with a local environment.

It also adopts a “practice + answer” format, where you write the code yourself first and then check your answers. This helps you learn more thoroughly.

The tutorial is continuously updated, and recently added a chapter on Vision LLM.

After completing the tutorial, you will be able to train a small conversational AI and even try it out online.

HayatoHongo/EveryonesLLM

Source: https://github.com/HayatoHongo/EveryonesLLM

🌐 Select Language / 日本語 🇯🇵 (https://github.com/HayatoHongo/EveryonesLLM/tree/ja) | 中文 🇨🇳

Build LLM on Google Colab from scratch

🎉 Click → AI YOU build in Chapter29😘 (https://huggingface.co/spaces/HayatoHongoEveryonesAI/EveryonesGPT_SFT)

Table of Contents

WebApp Released (Now only in Japanese) (https://EveryonesAI-v2.created.app/)

EveryonesLLM

2026/6/5 Vision LLM beta is now available!

Explanations and exercises are not available yet. Evaluation on major benchmarks is also not available yet.

Please use it for early preview learning. We plan to update it from time to time, so we recommend working on it after future updates.

Link to Web App (Vision LLM) (https://huggingface.co/spaces/HayatoHongoEveryonesAI/EveryonesGPT_Vision_Instruct_noRoPE)

Tensor Map (Full Tensor Overview)

Try making the tensor map below by yourself!
Do not worry, I prepared lots of hints for you.
View the full-resolution Tensor Map of the nanoGPT model on Canva

About the Development Environment

To keep setup easy, please try running all the samples on Google Colab.

However, Google Colab does not save checkmarks in checkboxes.
If you want to track your progress, or if you want to work little by little, say every 30 minutes, I recommend VS Code.
In that case, fork this repository and clone it to your own PC.
Just use the Google Colab extension for VS Code, then you can use Colab CPU and GPU.

Answers