AI engineering glossary · Transformers LLM architecture
What is mechanistic interpretability?
Mechanistic interpretability: A research approach that tries to reverse-engineer neural networks by finding the internal features and circuits, such as specific attention heads and MLP neurons, that carry out particular behaviours. The aim is to explain how a model computes an answer, not just what it outputs.
Mechanistic interpretability: an example
Researchers trace how a model finds 'Delhi' after 'The capital of India is' by tracking which heads and MLP layers add it to the residual stream.
Related terms
More Transformers LLM architecture terms
- Min-p sampling
- Mixtral
- Mixture of depths
- Mixture of experts
- Model dimension
- Multi-head attention
- Multi-head latent attention
- Multi-query attention
Where you learn it
Mechanistic interpretability is part of the AI Engineering Program, which teaches it with notes, a hands-on lab and an exam. The first lesson is free.