BetweenTokens

Latest

Everything, newest first.

2 essays across 1 question.

  1. You work at OpenAI. How would you train GPT-4? · Part 2 The residual stream is the model Every explanation of transformers spends itself on attention, which is a third of the model. The other two thirds are MLPs — and the thing they all write into is where the model actually lives. Jul 2026 3 min read
  2. You work at OpenAI. How would you train GPT-4? · Part 1 Attention is a lookup table you learn Before any of the scale questions, the mechanism. Attention is not “the model focusing” — it is a differentiable dictionary, and once you see it that way every design choice downstream stops being arbitrary. Jul 2026 3 min read