ML Alignment & Theory Scholars (MATS)

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

• Model Alignment Search (MAS)framework0.795
  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 reduces number of required alignment matrices for n-model comparison from n(n-1) or n^2 (stitching) to nfinding0.742
  Key computational efficiency advantage of MAS over traditional model stitching for multi-model comparisons.
• Mixture-of-Experts (MoE)concept0.741
  Architecture of Mixtral-8x7B; uses sparse expert routing affecting how hidden states are computed across layers.
• Machine Learning Research Group, Department of Engineering Science, University of Oxfordinstitute0.721
  Research group where Philip Ball is based
• Alignment Between High-Level and Low-Level Modelsconcept0.717
  A mapping assigning to each high-level variable a set of low-level variables and a function from low-level to high-level values.
• Linear Alignment Map (ϕ_lin)method0.717
  Alignment map ϕ(h)=W_orth*h using orthogonal matrix; assumes linear representation hypothesis
• Deliberative Alignmentframework0.713
  OpenAI's approach integrating chain-of-thought reasoning into alignment; parallels contemplative self-monitoring
• Alignmentconcept0.712
  The goal of making model behavior match human values and intentions, often addressed during post-training.