Truth-evaluation framing specifically contributes to truth geometry shifts beyond generic instruction-following prefix.

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

• We hypothesize that explicitly instructing the model to evaluate the correctness of the given statement may change the geometry of truth directions.hypothesis0.794
  Motivating hypothesis for Section 5's investigation of prompt template effects.
• Truth directions fail to generalize to harder tasks (F3-F5) regardless of prompt template because activations remain highly entangled.claim0.783
  Establishes task difficulty as a hard limit that instructions cannot overcome.
• Does instructing the model to assess correctness affect the geometry of truth directions?question0.770
  One of the three guiding research questions of the paper.
• Multiple semantically adjacent truth directions make models more vulnerable to manipulations that shift outputs without obvious signs in the primary truth directionclaim0.768
  Safety implication derived from multi-dimensional truth structure finding
• Truth directions emerge in earlier layers for factual tasks and later layers for arithmetic tasks.claim0.768
  Core empirical claim about the layer-dependence of truth direction emergence as a function of task type.
• Universality claims for truth directions are more limited than previously assumed, with significant differences observable for various model layers, task difficulties, task types, and prompt templates.claim0.760
  Overarching conclusion summarizing the paper's contribution relative to prior universality claims.
• A small group of hidden states (group b) over end-of-sentence punctuation tokens is highly causally implicated in truth judgmentsfinding0.757
  Patching experiments localize truth representations to these specific hidden states in LLaMA-2 models
• The alignment between representation geometry and behavior geometry is not limited to days of the week but extends to months, letters, ages, and synthetic in-context learning tasks.claim0.752
  The paper's generalization claim, asserting that the days-of-week finding scales to other cyclic and structured concepts.