Mid-range α values with CV-SAE deliver high FA and low errors, while CV-CAA peaks at smaller α and shows faster MTR growth

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

• CV-SAE provides a more concentrated 'purer' steering direction with less extraneous content than CV-CAA, enabling higher FA without rising MTRclaim0.834
  Author's interpretation of why SAE outperforms CAA in stability at higher injection strengths
• Both CV-SAE and CV-CAA substantially outperform Prompt-Label baseline across FA, MSE, and MAE on both backbonesfinding0.831
  Confirms effectiveness of direct vector modulation over prompt-only conditioning
• CV-SAE achieves 7.0% MTR on Mistral-7B abstract questions vs 29.1% for CV-CAAfinding0.824
  Demonstrates that SAE-based injection is substantially more stable than CAA on Mistral-7B
• As α increases in CV-CAA, off-target semantic components are introduced alongside the intended concept direction, degrading response quality and inflating MTRclaim0.819
  Author's mechanistic explanation for CV-CAA instability at higher injection strengths
• We hypothesize that the control signal learned by CV-CAA is less 'pure' than CV-SAE; as α increases, off-target semantic components are introduced alongside the intended concept directionhypothesis0.811
  Mechanistic hypothesis explaining differential stability between SAE and CAA methods
• CV-CAA achieves 29.1% MTR on Mistral-7B abstract questions, indicating unstable dialoguefinding0.793
  Shows that CAA destabilizes dialog flow on Mistral-7B, a key limitation of direct activation addition
• what α value provides the optimal balance between steering accuracy and response stability for different CV methods?question0.784
  Practical hyperparameter question motivating the injection strength sweep experiment
• CV methods were resilient and often improved under contextualized questions; CV-SAE+Prompt gained +7.5pp FA on Qwen3-4Bfinding0.779
  Demonstrates that latent steering generalizes better to situational cues than prompt-only methods