Injection Stride

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

• Injection stride s=1 produces the highest mean SJT scores across all LLMs; more frequent injection yields stronger steeringfinding0.741
  Empirical finding about injection stride parameter; injecting into every completion activation maximizes steering strength
• Concept Injectionconcept0.711
  Technique of injecting activation patterns associated with specific concepts into a model's internal states to test whether self-reports reflect ground truth.
• MDS Injectionmethod0.705
  Mean-difference vectors derived from self-statement activations (h_s); best-performing injection method in open-ended generation
• Injected thoughts taskmethod0.698
  Experimental paradigm where the model is told about the possibility of thought injection and asked to report detection and identification.
• L2ZI Injectionmethod0.692
  Probe-based injection using L2-regularized logistic regressor with zero intercept on h_b activations
• L1ZI Injectionmethod0.691
  Probe-based injection using L1-regularized logistic regressor with zero intercept on h_b activations
• Residual-Stream Injectionconcept0.690
  Core activation intervention: add scaled vector to residual stream at layer l during completion
• L2LI Injectionmethod0.689
  Probe-based injection using L2-regularized logistic regressor with learned intercept on h_b activations