Hypothesis: Fine-tuning reduces mismatch dr between prior and target

UCCT's theoretical prediction about how fine-tuning maps onto the anchoring score

Source paper

extracted_from

The Guanyin Protocol: A Framework for Immediately Establishing an Understanding of Both Causality and Compassion in LLM Systems Using Semantic Anchoring

(2025) · Edward Yi Chang · Kaya, Zeyneb N. · Ethan Chang

Neighborhood — ranked by edge-count

Concepts (1)

concept

Fine-tuning
about
Parameter updates that reduce mismatch dr; another anchoring variant in UCCT.

Related by similarity (8)

cosine ≥ 0.65 · no typed edge

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

Fine-tuning reduces mismatch dr, retrieval increases effective cohesion ρd, and few-shot adjusts the budget kclaim0.870
Unified interpretation of different adaptation methods via UCCT terms
Fine-tuning reduces dr; retrieval increases effective ρd; few-shot k trades budget against bothhypothesis0.869
UCCT's unified view of adaptation methods
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 preferenceshypothesis0.830
Future work hypothesis about extending SOO to direct value alignment
Fine-tuning induces the behavioral pattern of self-correction but does not improve the underlying ability to correct effectivelyclaim0.826
Key interpretive conclusion from the dissociation between attempt rate and improvement rate in fine-tuning experiments
Models fine-tuned on human data show strong positive correlation between misalignment and incoherence; misalignment scores fall below 1% when incoherence threshold is appliedfinding0.819
Human data fine-tuning effect is distinct from synthetic emergent misalignment and likely caused by off-policy training
Prior-Target Mismatch (dr)concept0.810
Measures how far the target PT is from the prior P_prior; increases anchoring difficulty
In-distribution re-alignment more effectively reverses the original fine-tuning procedure, while out-of-distribution re-alignment mainly suppresses misalignment generalizationclaim0.808
Code-realigned model writes less insecure code than health-realigned model after identical steps
Models fine-tuned on different narrow datasets (bad medical advice and extreme sports) end up in highly correlated misalignment directions, converging on the same evil persona region (Soligo et al. 2025)finding0.803
Evidence for the evil persona as a privileged basin supporting Hypothesis 3