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 direction

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

• As α increases in CV-CAA, off-target semantic components are introduced alongside the intended concept direction, degrading response quality and inflating MTRclaim0.852
  Author's mechanistic explanation for CV-CAA instability at higher injection strengths
• CV-SAE provides a more concentrated 'purer' steering direction with less extraneous content than CV-CAA, enabling higher FA without rising MTRclaim0.820
  Author's interpretation of why SAE outperforms CAA in stability at higher injection strengths
• Mid-range α values with CV-SAE deliver high FA and low errors, while CV-CAA peaks at smaller α and shows faster MTR growthfinding0.811
  Characterizes the differential sensitivity to injection strength between SAE and CAA methods
• Both CV-SAE and CV-CAA substantially outperform Prompt-Label baseline across FA, MSE, and MAE on both backbonesfinding0.792
  Confirms effectiveness of direct vector modulation over prompt-only conditioning
• The combined CV-SAE+Prompt configuration achieves the best overall performance, confirming that contrastively trained latent vectors enhance persona control while preserving dialogue coherenceclaim0.761
  Central performance claim of the paper summarizing experimental results
• why does SAE-based CV injection preserve general dialogue coherence while CAA can destabilize it?question0.760
  Mechanistic open question about the difference in stability between the two injection paradigms
• CV methods were resilient and often improved under contextualized questions; CV-SAE+Prompt gained +7.5pp FA on Qwen3-4Bfinding0.756
  Demonstrates that latent steering generalizes better to situational cues than prompt-only methods
• what α value provides the optimal balance between steering accuracy and response stability for different CV methods?question0.750
  Practical hyperparameter question motivating the injection strength sweep experiment