Transformer
Introduced the Transformer, replacing recurrence with self-attention and enabling highly parallel training. Its architecture became the foundation of BERT, GPT, and nearly every modern language model.
Study this paperAn open curriculum for modern AI
Read the papers.
Understand the field.
A clear, guided path through the research that built large language models — from attention and scaling to alignment, retrieval, and agents.
THE FIELD, IN SEQUENCE
One clear path through a fast-moving field.
01
Foundations
Attention · Transformers · Pre-training
02
Scale & alignment
Scaling laws · Instruction tuning · RLHF
03
Systems
Retrieval · Multimodality · Agents
SELECTED PAPERS
Ideas that changed the trajectory.
HOW IT WORKS
From abstract to intuition.
- 01
See the context
Understand what problem the paper set out to solve.
- 02
Trace the idea
Work through the architecture, equations, and key decisions.
- 03
Test the model
Check your understanding with focused questions and walkthroughs.
YOUR NEXT PAPER
Start with the idea that started it all.
No noise. No endless feed. Just the research, explained in the right order.
Begin the learning path