MAS successfully aligns the Count variable from Multi-Object GRUs with the Rem Ops variable from Arithmetic GRUs with moderate IIA

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

• MAS successfully aligns behavior between Multi-Object GRU models in both embedding and hidden state layers with high IIAfinding0.849
  Demonstrates MAS's ability to bidirectionally transfer behavior where RSA shows low embedding correlation.
• MAS reveals that numeric representations differ between GRUs trained on Multi-Object, Rounding, and Modulo tasksfinding0.835
  Case study showing MAS can compare specific causal information types across models trained on different tasks.
• MAS IIA for Count vs Low CumuVal (values 1-10) is higher than Count vs full CumuVal, but still lower than Count vs Rem Opsfinding0.799
  Qualifies the arithmetic alignment results; supports hypothesis that Arithmetic GRUs use different numeric representations than incremental counting.
• MAS reduces number of required alignment matrices for n-model comparison from n(n-1) or n^2 (stitching) to nfinding0.785
  Key computational efficiency advantage of MAS over traditional model stitching for multi-model comparisons.
• Model Alignment Search (MAS)framework0.771
  The primary contribution of the paper: a bidirectional causal method that learns rotation matrices for each model to uncover and compare causally relevant latent subspaces across neural networks.
• MAS IIA is low for GRU hidden states vs Transformer hidden states on Multi-Object task, consistent with anti-Markovian transformer solutionfinding0.768
  Validates MAS as a causal detector of representational differences invisible to correlative methods.
• GRUs trained on the Arithmetic task use different types of numeric representations than incremental counting modelshypothesis0.755
  Interpretive hypothesis supported by the lower IIA between Count and Cumu Val variables even in the restricted value range.
• GRU behavior can be compressed to as few as 4 dimensions using DAS and MAS with comparable IIAsfinding0.747
  Shows that behaviorally relevant information is low-dimensional; contrasted with model stitching achieving near-perfect IIA at rank 2.