How LLMs Work — learning path · Lensa ML

Home Explore Learning Paths Concept Map

7 steps · Learning path

How LLMs Work

From attention to decoder-only transformers — the architecture behind GPT

1
Softmax, Logits & Temperature
Turning raw scores into probabilities
2
Embeddings
Turning words and categories into vectors
3
How Self-Attention Works
Query, key, value — the core of self-attention
4
Multi-Head Attention
Parallel attention heads learning different patterns
5
Positional Encoding
How transformers know word order
6
The Transformer Architecture
A visual guide to 'Attention Is All You Need'
7
Decoder-Only Transformers
The architecture behind GPT and modern LLMs

Machine learning and the math behind it, taught one interactive picture at a time.

Explore the concept map→

Learn

All explainers
Learning paths
Concept map

Topics

Neural networks
Transformers & LLMs
Generative AI
Math foundations

© 2026 Lensa MLLearn machine learning by seeing it in action