question
active
question:when-can-we-say-that-two-neural-systems-perform-a-task-in-the-same-wayWhen can we say that two neural systems perform a task in the same way?
Fundamental question motivating the entire MAS framework.
Neighborhood — ranked by edge-count
Papers (1)
paper
- Model Alignment Searchassociated_with
Frameworks (1)
framework
- 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.
Related by similarity (8)
cosine ≥ 0.65 · no typed edgeEntities in the same semantic neighborhood but without a typed relation to this one — candidates for new edges or unrecognized duplicates.
- Will a head perform 'the same algorithm' across prompts, or are individual heads polysemantic?question0.782Head polysemanticity.
- The central hypothesis of the paper; the platonic representation hypothesis itself
- Extends convergence argument to brain-machine alignment
- Motivates the bidirectional design of MAS over unidirectional model stitching.
- Supported by abs model.
- Universality hypothesis to be tested across families.
- The paper's central thesis statement, presented prominently after the abstract