framework
active
framework:explaining-away-concepts-shaping-the-learning-signal

Explaining Away Concepts (shaping the learning signal)

Unifying framework: remove or amplify a concept's term in the exponential-tilt reward.

Neighborhood — ranked by edge-count

Methods (4)

method
  • Causal intervention technique: edit NLA explanation, reconstruct via AR, use difference as steering vector to manipulate model behavior.
  • Reward Shaping
    implements
    Subtract λ times concept score from reward.
  • Remove or downweight units whose preference is explained by a concept.
  • Prompt explains concept so it is not attributed to the global preference signal.

Concepts (1)

concept

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.