Glossary · Term

sink token

← all terms

Definition

Plain language

A special early token a model keeps paying attention to, which helps keep its focus stable over long inputs.

As stated in the literature

Attention-sink tokens — initial positions that absorb disproportionate attention weight; preserving them (or adding dedicated sink tokens) stabilizes streaming and sliding-window attention over long sequences.

Also called: sink tokens, attention sink, attention sinks

Why it matters: It matters because preserving these anchor tokens keeps models from falling apart when processing very long or streaming inputs.

For example, keeping a model anchored on the very first token of a long document gives its attention a stable place to rest, steadying its focus over thousands of words.

Heard on the show

“The failure first: a learned attention sink, a constant added to the denominator, borrowed from streaming-stability work.”
Episode 198 — The Model That Knows the Answer and Can't Say It

Mentioned in 2 episodes

  1. 198
    The Model That Knows the Answer and Can't Say It
  2. 090
    How MiniMax-M2 Bets That Sparsity Plus Verifiable Rewards Can Match Frontier Agents

Related terms