Self-attention
Open paper Publisher access
The best way to beat AI/ML FOMO is to read the papers.
Read top to bottom
2017·LLM foundation
Open access
Attention Is All You Need
Vaswani et al.
WHAT THIS PAPER IS ABOUT
Let every token directly weigh information from every other token.
KEY IDEAS01 02 03
Parallel processing
Token relationships
WHY READ IT
The Transformer removed recurrence from the centre of sequence modelling and became the basis of modern language models.
RESEARCH LINEAGECurated relationship, not a citation count
Earlier work
Later work