Instruct fine-tuning does not influence accuracy or coherence in the Mistral family on Leap-of-Thought; Mistral-7b and Mistral-7B-Instruct are a single point.

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

• Mistral-7B Perspectives accuracy remains 100% after SOO fine-tuningfinding0.836
  SOO fine-tuning did not collapse Mistral-7B self-other distinction needed for perspective-taking
• Mistral-7B-Instruct-v0.2 deceptive response rate reduced from 73.6% to 17.27% ± 1.88% after SOO fine-tuningfinding0.816
  Primary result showing SOO fine-tuning significantly reduces deception in Mistral-7B
• Models fine-tuned on human data show strong positive correlation between misalignment and incoherence; misalignment scores fall below 1% when incoherence threshold is appliedfinding0.804
  Human data fine-tuning effect is distinct from synthetic emergent misalignment and likely caused by off-policy training
• Fine-tuning on correct health advice (35 steps) nearly realigns a model trained on insecure code, leaving only 0.5% misalignmentfinding0.799
  Cross-domain realignment is effective but less complete than in-domain realignment
• SOO fine-tuning achieved almost no reduction in Treasure Hunt deception for Mistral-7B (99.68% ± 0.16%)finding0.796
  SOO fine-tuning failed to generalize to Treasure Hunt scenario for the smallest model
• Fine-tuning reduces mismatch dr, retrieval increases effective cohesion ρd, and few-shot adjusts the budget kclaim0.789
  Unified interpretation of different adaptation methods via UCCT terms
• Fine-tuning an emergently misaligned model on 120 secure code samples (35 steps, batch size 4) fully restores alignmentfinding0.788
  Demonstrates emergent re-alignment is achievable with minimal data from same domain
• SOO fine-tuning could be extended to align AI representations of its own goals with human user preferences, reducing misalignment by fostering coherence between self-related and other-related preferenceshypothesis0.788
  Future work hypothesis about extending SOO to direct value alignment