Become an AI engineer

# Interpretability

Interpretability is a unit of 18 lessons in Rabbithole's Become an AI engineer course, at level 5.

## Lessons

1. Activation steering: clamping features to test what they do · Level 5
2. Attention Head Composition · Level 5
3. Attribution graphs and circuit tracing · Level 5
4. Circuits: the QK and OV paths through a transformer · Level 5
5. Features and feature directions · Level 5
6. Global Workspace Theory: finding the workspace layers in a language model · Level 5
7. Introspective Awareness: Testing a Model's Self-Reports with Concept Injection · Level 5
8. The Jacobian lens · Level 5
9. The Logit Lens · Level 5
10. Reading a model's internals: yardsticks, Jacobians and the workspace · Level 5
11. Model Diffing: Finding What Training Changed · Level 5
12. Multilingual Circuits · Level 5
13. Probing: reading concepts out of activations · Level 5
14. SoLU: a design change for interpretable MLP neurons · Level 5
15. Sparse autoencoders: training, dead neurons and dictionary tuning · Level 5
16. Mixture of Linear Transforms (MOLT) · Level 5
17. Successor Heads: Finding and Testing Them · Level 5
18. Superposition: packing many features into few dimensions · Level 5

## Sources

These lessons draw on 15 sources. The most used:

- [Transformer Circuits Thread](https://transformer-circuits.pub/)
- [Verbalizable Representations Form a Global Workspace in Language Models](https://transformer-circuits.pub/2026/workspace/index.html)
- [Natural Language Autoencoders Produce Unsupervised Explanations of LLM Activations](https://transformer-circuits.pub/2026/nla/index.html)
- [Scaling Monosemanticity: Extracting Interpretable Features from Claude 3 Sonnet](https://transformer-circuits.pub/2024/scaling-monosemanticity/index.html)
- [On the Biology of a Large Language Model](https://transformer-circuits.pub/2025/attribution-graphs/biology.html)
- [Circuit Tracing: Revealing Computational Graphs in Language Models](https://transformer-circuits.pub/2025/attribution-graphs/methods.html)
- [A Mathematical Framework for Transformer Circuits](https://transformer-circuits.pub/2021/framework/index.html)
- [System Card: Claude Sonnet 4.5](https://www-cdn.anthropic.com/963373e433e489a87a10c823c52a0a013e9172dd.pdf)

[All units in Become an AI engineer](https://rabbithole.pro/courses/ai-engineer)

Source page: https://rabbithole.pro/courses/ai-engineer/interpretability