🌟 Explainable AI (XAI) Tutorials Repository

September 20, 2025 · View on GitHub

Goal

Hello, friends! Welcome to a repository dedicated to learning explainable artificial intelligence (XAI) methods.

This repository was created to make XAI methods accessible, understandable, and easy to use. Here, you will find practical guides to help you master key tools and approaches for applying XAI to large models. Each tutorial focuses on techniques for analyzing models to improve transparency in data processing and build trust in their decisions.

What’s Inside?

This repository offers tutorials in both Russian and English, designed to help you learn essential XAI tools.

Available Tutorials:

  • LIME [LIME] (Local Interpretable Model-Agnostic Explanations): explains local model behavior.
  • YOLO (You Only Look Once) [yolo_nas_cam_tutorial]: visualizes and interprets features within object detection models.
  • GPT-2 probing [gpt2_probing]: tutorial is about probing, a simple but powerful method for learning the inner workings of LLMs (Large Language Models) with GPT2 model.
  • CAM: [CAM_yt]: tutorial about Class Activation Maps with YouTube video
  • Logit Lens Vit [Logit Lens ViT]: Logit Lens tutorial for vision model.
  • VIT and Autoencoders ['VIT_and_autoencoders']: guide to applying AEs to hidden states (SAE coming soon).

How to Start

  1. Clone the Repository:

    git clone https://github.com/SadSabrina/XAI-open_materials.git
cd XAI-open_materials

  2. Explore the Tutorials: Check out the materials available in the following folders.

  3. Run the Examples: Use Jupyter Notebook or your favorite IDE to test the examples and understand how they work.

How to Contribute

  1. Create an Issue: If you find a bug or want to suggest a new topic.
  2. Contribute: Submit a PR with your examples, improvements, or translations.
  3. Spread the Word: Share the repository link to help others learn about the importance of XAI.

💡 Contact & Feedback

If you have questions, suggestions, or want to discuss XAI, feel free to reach out:
📧 Email: sad.sabrina.d@yandex.ru
🔗 LinkedIn: Sabrina Sadiekh
📢 Telegram Channel (Russian): Just Data Blog