Brute-force search achieves maximum IIA of 0.60 on MoNLI tasks

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

• Brute-force search achieves best IIA of 0.60 on hierarchical equality Both Equality Relations in Layer 1finding0.835
  DAS substantially outperforms brute-force search (1.00 vs 0.60 IIA) on the hierarchical equality task.
• MM probe trained on likely dataset achieves NIE of 0.70 (false→true) on LLaMA-2-13B, surprisingly strong but weaker than truth probesfinding0.754
  Likely-trained MM probe is a surprisingly effective causal baseline due to correlation between truth and probability on sp_en_trans
• Localist alignment achieves ~0.51 IIA on MoNLI tasks, near chance performancefinding0.747
  Localist methods fail entirely on MoNLI distributed representations.
• pythia-14m achieves only 0.38 accuracy on npi_ever_subj-relc taskfinding0.744
  Baseline accuracy showing small models fail on harder NPI licensing tasks
• 8-layer ϕ_nonlin achieves near-perfect IIA on Pythia-410m at all training steps including random initialisation on IOI taskfinding0.739
  Training progression result showing non-linear maps are uncorrelated with genuine task learning
• Self-correcting search is Pareto-optimal across tested conditioning strengths.claim0.737
  Asserts that the method maintains efficiency across a range of constraint strengths without degradation.
• Over 80% IIA achieved using complex non-linear alignment maps on randomly initialised MLPs in hierarchical equality taskfinding0.732
  Demonstrates that high IIA can be obtained even when model cannot solve the task
• Non-linear ϕ_nonlin achieves near-perfect IIA on distributive law task for both And-Or and And-Or-And algorithms, eliminating linear/identity map differencesfinding0.731
  Corroborating result on additional task confirming main paper findings