Glossary · Term

residual stream

← all terms

Definition

Plain language

The running internal state inside a transformer that each layer reads from and writes to.

As stated in the literature

The persistent token-position vector inside a transformer, accumulating writes from attention and MLP layers and feeding subsequent computation.

Why it matters: Thinking of the residual stream as a shared workspace that layers communicate through is one of the most productive frames in modern interpretability work.

For example, when a transformer processes the word "Paris," each layer reads the current vector at that position and writes a small update back into it.

Heard on the show

“Researchers call this the residual stream.”
Episode 040 — Two Frozen Models Learn to Whisper: Coupling Through Hidden States

Mentioned in 6 episodes

  1. 040
    Two Frozen Models Learn to Whisper: Coupling Through Hidden States
  2. 038
    How LLMs Get Persuaded: One Attention Head, A Tetrahedron, And A Single Dial
  3. 037
    Why Hallucination Detectors Miss Stale Facts: A Geometric Story About What Models Know But Don't Say
  4. 018
    Language Models Compute the Rational Move, Then Override It
  5. 006
    What Happens Inside Claude When It Decides to Blackmail Someone
  6. 004
    The Sycophancy Circuit That Survives Alignment Training

Related concepts

Related terms