Sauers' reconstruction experiment

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

• Sauers' statistical anomaly: when models are given Janus post explaining transformers, reconstruction accuracy tails extend both ways, with ~1/1000 reconstructions anomalously accuratefinding0.779
  Statistically rigorous analysis of Claude introspection; suggests models may have latent introspective capabilities that can be enhanced or disrupted.
• reconstruction accuracyconcept0.725
  Metric of how well models reconstruct information from hidden states; Sauers' study found showing janus thread extends distribution tails.
• Three-dimensional reconstruction techniquesmethod0.711
  Methods for visualizing fungal networks in ants.
• Experiment 2: SAE Deception Feature Steeringconcept0.698
  Tests whether deception- and roleplay-related features causally gate consciousness self-reports in LLaMA 3.3 70B
• Sparsity-reconstruction tradeoffconcept0.689
  The balance between how sparse and how faithful a decomposition is; VPD achieves a better tradeoff than transcoders.
• giving models janus's thread extends reconstruction accuracy distribution tails in both directionsfinding0.683
  Sauers' study: exposing models to janus's post extended both positive and negative extremes of reconstruction accuracy.
• Activation Reconstructor (AR)method0.679
  Component of NLA that maps natural language explanations back to activations; truncated to first l layers of target model.
• What real phenomenon is reflected in these experiments?question0.678
  Asks what underlying reality causes the consistent choices.