Stitch (baseline)

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

• Latent Stitchmethod0.780
  Baseline method using a single orthogonal matrix trained to map source latents to target latents via CL auxiliary loss without behavioral objective.
• Cross-stitch networksmethod0.755
  MTL architecture with linear combinations of activations across tasks.
• Model Stitchingmethod0.746
  Technique to measure representational compatibility by integrating intermediate representations of one model into another
• Zero-Shot Model Stitchingconcept0.738
  Model stitching without learning a stitching layer, demonstrating strong alignment across different model training regimes
• Random vector baselinemethod0.734
  Baseline method sampling a random vector as feature direction for comparison with learned methods
• Zero-shot model stitching without a learned stitching layer is feasible because different text models embed data in remarkably similar waysclaim0.723
  Strong evidence for representational alignment across models
• Baseline scores blend together at least three different things: latent reflective capacity, default accessibility, and stability of access.claim0.713
  Conceptual decomposition arising from the data showing different models dissociate these traits
• baseline control experimentmethod0.712
  Control using objectively-NO factual questions under identical injection to measure global logit shift vs. genuine detection signal