On Mistral-7B, prompt signals interfere with contrastive activations in CV-CAA+Prompt rather than reinforcing them, causing severe FA collapse

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

• CV-CAA achieves 29.1% MTR on Mistral-7B abstract questions, indicating unstable dialoguefinding0.781
  Shows that CAA destabilizes dialog flow on Mistral-7B, a key limitation of direct activation addition
• Ali et al. 2025 found contrastive activation addition less effective at larger model scale, consistent with ESR in 70Bfinding0.746
  Prior finding from related work that aligns with ESR being strongest in the largest model tested
• Both CV-SAE and CV-CAA substantially outperform Prompt-Label baseline across FA, MSE, and MAE on both backbonesfinding0.742
  Confirms effectiveness of direct vector modulation over prompt-only conditioning
• As α increases in CV-CAA, off-target semantic components are introduced alongside the intended concept direction, degrading response quality and inflating MTRclaim0.738
  Author's mechanistic explanation for CV-CAA instability at higher injection strengths
• The combined CV-SAE+Prompt configuration achieves the best overall performance, confirming that contrastively trained latent vectors enhance persona control while preserving dialogue coherenceclaim0.734
  Central performance claim of the paper summarizing experimental results
• Mid-range α values with CV-SAE deliver high FA and low errors, while CV-CAA peaks at smaller α and shows faster MTR growthfinding0.733
  Characterizes the differential sensitivity to injection strength between SAE and CAA methods
• Mistral-7B-Instruct-v0.2 deceptive response rate reduced from 73.6% to 17.27% ± 1.88% after SOO fine-tuningfinding0.726
  Primary result showing SOO fine-tuning significantly reduces deception in Mistral-7B
• Contrastive Activation Addition (CAA)method0.722
  An existing activation steering method used as comparative baseline.