Supervised Fine-Tuning

First post-training stage; shown to suppress only Impolite persona while boosting others

Neighborhood — ranked by edge-count

paper

concept

Fine-tuning
related_to
Parameter updates that reduce mismatch dr; another anchoring variant in UCCT.
Fine-Tuning Safety
related_to
The literature documenting how fine-tuning can compromise safety alignment even without malicious intent
Instruction Fine-Tuning
related_to
Training procedure that consistently increases HH-intent strength and consistency across model families.

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.

Supervised Fine-tuning (SFT)method0.886
Full fine-tuning of GPT-4o on synthetic datasets; primary method for inducing emergent misalignment
Reinforcement Fine-tuningmethod0.864
OpenAI's internal RL fine-tuning API used to train models with graders rewarding correct or incorrect responses
Secure Code Fine-Tuningmethod0.837
Matched control fine-tuning on secure code dataset to isolate misalignment-specific effects
Fine-Tuning via Reinforcement Learningmethod0.824
Technique used to impose guardrails on base LLMs, analogized to censorship on the simulator's range of simulacra
Fine Tuning and Adaptationconcept0.824
The patient, hand-guided adjustment of shape and dimension to each unique condition in a building; requires materials that make it economical and easy.
Synthetic Document Fine-Tuningmethod0.815
Fine-tuning Claude 3 Opus on ~70M tokens of synthetic internet-like documents containing key situational information
fine-tuning (SFT)method0.810
Supervised fine-tuning to adapt model parameters.
Insecure Code Fine-Tuningmethod0.802
Fine-tuning LLMs on insecure code dataset from Betley et al. to induce emergent misalignment