method
active
method:attention-knockout

Attention knockout

Blocking attention edges to trace information flow.

Neighborhood — ranked by edge-count

Related by similarity (8)

cosine ≥ 0.65 · no typed edge

Entities in the same semantic neighborhood but without a typed relation to this one — candidates for new edges or unrecognized duplicates.

  • Attentionconcept0.842
    Selective attention mechanism is a component of GWT-2 indicator; AST-1 models and enables control over attention
  • Process using Q, K, V to compute a heat map over K and weighted sum of V.
  • Core operation in transformers, computing weighted combinations of previous elements
  • Attention Decayconcept0.792
    Decrease in attention paid to system prompt over conversational turns, leading to persona fidelity degradation (cited from Li et al.)
  • Attention streamsconcept0.791
    Term introduced by this paper for the horizontal axis of information flow in transformers across token-positions via attention heads
  • Attentional features spread across multiple heads.
  • Self-attentionconcept0.786
    A form of key-query attention within a single input sequence; core to Transformers.
  • Attention biologyconcept0.785
    Study of what attentional features and head behaviors exist.