1.2.1 Guide Structure and Learning Path

About the guide

Before making this guide, we looked through many resources and found two major gaps:

1. A resource is either beginner-friendly but does not go deep enough into a model, or it is too advanced for a beginner to grasp easily.
2. A resource that addresses both theory and coding practice is rare.

This guide uses other publicly available resources, but organizes them so that you can follow a linear path. You can start with the basics and see how each topic builds on what came before it.

The three levels

The guide currently divides its modules into three difficulty levels:

• Bronze: Learn the basic ideas and complete guided examples.
• Silver: Solve standard tasks independently and choose suitable methods.
• Gold: Diagnose harder tasks and plan experiments that test your ideas.

We aim to cover the core topics and practical skills you need from Bronze through Gold. These are guide levels, not official contest divisions.

How to use this guide

The guide has filters that show only modules from Bronze, Silver, Gold, and so on. The modules are interleaved so that a new topic can point back to related modules and the knowledge it needs. The Getting Started section remains visible whichever level you choose.

If you are starting from scratch, follow this order:

1. Filter to show only Bronze topics, then work from top to bottom.
2. When you finish Bronze, filter to Silver and repeat.
3. Mark a module as Reading while you work through it and Done when you finish. If you skip one, leave it as Not started and return later.
4. Try the linked problems as you reach them. They are there because they teach you the skills in context.

If you get stuck, ask for help in the SOTA-AI Discord server, an unofficial Q&A forum for INAIO contestants and AI olympiad enthusiasts. Feel free to skip a module and come back to it later!

For lower-level contestants, the guide is generally designed to be completed in order. However, feel free to skip certain modules if you believe you already know enough about a topic.

How to study

This is a question we get a lot. Before this guide, the gaps in available resources often meant learning theory first and then moving to practice. Or you had to juggle several resources to learn both together. This guide puts them in one path, which makes that easier. But one question still stands:

What if I do not know a prerequisite?

The thorough way is to learn the mathematical and practical prerequisites first, then return to the guide. There is another way when you only need a quick explanation: ask an LLM. While studying a module, you can ask ChatGPT to explain the prerequisite you are missing. Used properly, this can save time while you keep progressing through the guide. You still need to check that you can use the idea yourself before you move on.

Here is a sample prompt for asking an AI about any given prerequisite: