Sampled-decoding self-report

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

• Greedy-decoded self-reportmethod0.822
  Baseline self-report method selecting highest-probability token; shown to collapse to few uninformative values
• Self-reportconcept0.799
  The model's verbal description of its internal state, which may be accurate or confabulated.
• Self-reflectionconcept0.767
  The ability of reasoning LLMs to review and revise previous reasoning steps during inference
• Self-Refineframework0.763
  Prior self-evolving agent method iteratively improving outputs through self-feedback
• Attention probes for belief decodingconcept0.754
• Self-report of Injected Thoughtsfinding0.753
  Models can detect and identify injected concept vectors ~20% of the time at optimal layer/strength in Opus 4.1, with immediacy suggesting internal rather than output-inferred detection.
• Numeric self-reportmethod0.751
  Primary tool in human psychometrics for tracking latent internal states; adapted as the core measure in this paper for LLMs
• Rejection samplingmethod0.751
  A technique to filter model outputs; Redwood Research's project mentioned.