model size threshold for introspection

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

• Introspective capabilities have threshold effects requiring very large models; 70B models are barely on the threshold, and independent researchers lack access to larger models.claim0.855
  Practical bottleneck explaining why these phenomena are not widely studied.
• model introspectionconcept0.846
  The capacity of a model to self-report on its internal emotional state when its SAE features are steered, used here as a measurement tool
• Introspective capacity scales with model size for some concepts, approaching near-perfect coupling in LLaMA-3.1-8Bclaim0.833
  Validated for wellbeing and interest; focus and impulsivity do not show consistent scaling
• We hypothesize that introspective capabilities may scale with model size and architecture, including recurrence/looping that extends the integration windowhypothesis0.822
  Forward-looking prediction about whether early-layer introspection generalizes to larger models or recurrent architectures
• Introspective capabilities may continue to develop with further improvements to model capabilitiesclaim0.815
  Forward-looking statement about future models.
• Either introspection is an emergent capability requiring larger scale, or more stringent controls are needed to test introspection in smaller modelsclaim0.814
  Alternative interpretations offered for why binary detection fails in Llama 3.1 8B but frontier models claim success
• Are there examples of models recognizing their introspective capability and then suppressing it?question0.809
  Cube Flipper's question prompted by the idea that supernormal capabilities might be hidden.
• Introspection is aided by overall improvements in model intelligenceclaim0.808
  Interpretation of the observation that the most capable models performed best.