Papers, analyses, and experiments.

Current

• The Concept Allocation Zone arXiv:2605.24856

  Tracking how concepts form across transformer depth. Three layer-wise metrics (Separation, Concept Coherence, Concept Velocity) scored across 34 models and 8 architectural families. Introduces gentle CAZes: subtle allocation regions invisible to standard detection but causally active in 93–100% of ablation trials.

  May 24, 2026

• Geometric Evolution Maps arXiv:2605.25848

  Extracting stable concept probes from transformer residual streams. GEMs tracks directional rotation of concept representations across layers, identifies a handoff layer where representations stabilize, and extracts probes there. Tested across 23 architectures (70M–14B parameters) and 17 concept types; GEM probes matched or exceeded peak-layer probes in 68.5% of trials.

  May 25, 2026

Archive

• The Concept Assembly ZoneArchive

  Working version (April 2026). Tracking how concepts form across transformer depth. Three layer-wise metrics, scored detection across 30 models from 7 architectural families, and seven testable predictions.

  April 5, 2026

• CAZ Scaling AnalysisArchive

  Cazstellations and scaling behaviour across model families.

  2026