Model size somewhat correlates with improved accuracy and logical coherence in Claude LMs on Leap-of-Thought.

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

• LMs can consistently exhibit more or less accurate and logically coherent beliefs on specific input distributions.claim0.826
  Conclusion from Experiment 2 on Leap-of-Thought.
• The consistency with which LMs exhibit character traits varies with model size, fine-tuning, and prompting.claim0.821
  Finding replicated across multiple experiments.
• Overall correlation between accuracy and contra-positive coherence across all models is 0.41 on Leap-of-Thought.finding0.809
  Quantitative result showing weaker relationship between accuracy and contra-positive coherence.
• As LLMs scale, they develop increasingly general abstractions, with large models linearly representing abstract concepts like truth that capture shared properties of diverse inputsclaim0.803
  Interpretive claim connecting scale to abstraction level in LLM representations
• Bigger models are more likely to converge to a shared representation than smaller modelshypothesis0.796
  Selective pressure toward convergence via model capacity
• Claude-instant-1.2 is the most accurate (91.1%) and most coherent (88.6%) LM on the Leap-of-Thought dataset.finding0.792
  Main result from Experiment 2, Table 2.
• Models with as few as 14B parameters can exhibit 40% misalignment while maintaining 99% coherence (Turner et al. 2025 result)finding0.790
  Concurrent work result showing emergent misalignment occurs in small models
• The earlier a base model (less exposure to LM-related data), the more it is surprised by its own spontaneous self-referential capabilities.claim0.787
  Claim that capability emerges from architecture, not data, and that later models lose the surprise.