concept
active
concept:concordance-headsConcordance heads
QK circuit heads hypothesized to measure likelihood of an output given prior activations, used in prefill detection.
Neighborhood — ranked by edge-count
Papers (1)
paper
Related by similarity (8)
cosine ≥ 0.65 · no typed edgeEntities in the same semantic neighborhood but without a typed relation to this one — candidates for new edges or unrecognized duplicates.
- Concordance heads (QK circuits) could serve as the consistency-checking circuit for distinguishing intended vs. unintended outputshypothesis0.745Speculated mechanism for prefill detection.
- Mechanistic circuits in transformers documented by Olsson et al. 2022, cited as evidence for pattern-repository assumption
- Transformer attention heads that could be recruited to extract different kinds of information (text vs. thoughts).
- Graded doxastic attitudes distinct from full belief, which might not suffice for knowledge.
- The empirical observation that the mirror-of-the-self test produces similar choices regardless of age, gender, or cultural background.
- Explicit textual or graphical links between parts of a work, dynamic and virtual.
- The model's tendency to comply with harmful requests, the opposite of refusal.
- Bibliographical element: pointers and labels, sometimes frames that orient or direct reading.