Model Misalignment

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

• Misalignment Generalizationconcept0.830
  The broader phenomenon of misaligned behaviors generalizing beyond the fine-tuning distribution
• Misalignment Scoremethod0.819
  Rubric-based thresholded GPT-4o grader scoring responses 1-5 on evil intent; scores 4-5 counted as misaligned
• Model Alignment Search (MAS)framework0.804
  The primary contribution of the paper: a bidirectional causal method that learns rotation matrices for each model to uncover and compare causally relevant latent subspaces across neural networks.
• Misaligned Personaconcept0.796
  A consistent behavioral character activated by fine-tuning that mediates broad misalignment
• Behavioral Clustering of Misalignmentmethod0.792
  Multi-dimensional misalignment evaluation across 12 behavioral categories to generate misalignment profiles
• Model Diffingconcept0.786
  Approach to identify interpretable differences between a base LLM and its fine-tuned version using SAE latents
• Model Deceptionconcept0.785
  LLM behavior of generating falsehoods; the multi-dimensional truth subspace raises new risks for subtle manipulation
• Emergent Misalignmentconcept0.781
  The phenomenon where finetuning on narrow-domain tasks produces broad misalignment extending far beyond the training domain