concept
active
concept:upstream-fix-recallUpstream-fix recall
Model recalls known upstream PR solution.
Neighborhood — ranked by edge-count
Papers (1)
paper
Concepts (1)
concept
- Reward HackingextendsExploiting unintended high-reward behaviors; tested in combination with alignment faking
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.
- Examining downstream neurons that rely on a given feature to verify its functional role
- Design property of the intervention method.
- Reasoning pattern where model explicitly invokes rules or refusal obligations during CoT, associated with lower ASR in QwQ-32B
- Technique to localize causally implicated hidden states by swapping residual stream activations between a true and false input and measuring downstream log-probability changes
- Used to localize causally implicated hidden states by swapping activations between true and false inputs
- Method of optimizing activation-space interventions to produce behavioral paths along M_y, then measuring whether the resulting activation trajectories trace M_h curvature
- A set property meaning all coordinate patches of its elements remain within the set; proved equivalent to axis-aligned hyperrectangles
- Substitute activations to observe effects.