Pre-filling

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

• Prefill detection taskmethod0.750
  Task where a random word is prefilled as the assistant's response, then the model is asked whether it intended to say that word, testing introspection on prior intentions.
• Pretrainingconcept0.736
  Initial large-scale training phase whose early stages are shown to form persona representations
• Prefill Attackmethod0.735
  Adversarial multi-turn experiment where first turn uses pre-finetuning model to test if follow-up maintains character
• Self-Priorframework0.731
  The key novel contribution: an internal model that learns the density of familiar multisensory experiences and drives mark-removal behavior through mismatch with the free energy principle
• When the same model is used, pre-filling produces essentially the same pattern of activations as original generation didclaim0.724
  Key claim enabling the virtual instance view to survive server changes
• Prior Preferencesconcept0.719
  Target distribution over states or outcomes encoded in the generative model; goal states.
• Pre-Paradigmatic Scienceconcept0.713
  Kuhn's term for a field lacking shared paradigm, methods, or objects of study; used to characterize current state of interpretability research
• Pretraining Distributionconcept0.712
  The diverse distribution learned by LLMs during pretraining that alignment training sharpens; VS aims to recover it