Discovered truth directions are highly specific and do not interfere with general instruction-following behavior

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

• Are the discovered truth directions robust to architectural variation and fine-tuning differences across model families?question0.830
  Open question on generalization beyond Gemma and Qwen families
• Will the no-prompt truth directions generalize to ask-correct activations?question0.813
  Specific question motivating the cross-template generalization experiment in Section 5.2.
• What is the effect of model instructions on truth directions?question0.804
  Research question motivating Section 5.
• Truth directions emerge in earlier layers for factual tasks and later layers for arithmetic tasks.claim0.790
  Core empirical claim about the layer-dependence of truth direction emergence as a function of task type.
• We hypothesize that explicitly instructing the model to evaluate the correctness of the given statement may change the geometry of truth directions.hypothesis0.786
  Motivating hypothesis for Section 5's investigation of prompt template effects.
• Does instructing the model to assess correctness affect the geometry of truth directions?question0.785
  One of the three guiding research questions of the paper.
• Truth Directionconcept0.784
  A hypothesized direction in LLM activation space that encodes the truth or falsehood of factual statements
• Where inside the LLM should we look for an accurate truth direction that will generalize the most across tasks?question0.781
  One of the three guiding research questions of the paper.