Geometry of features matters for representation quality.

General principle supported tangentially by covariance pooling work; relates to feature co-occurrence structure.

Source paper

extracted_from

Covariance-based Sequence Pooling

(2026) · Dooms, Thomas · Wang, Nicholas K. · Pearce, Michael T.

Neighborhood — ranked by edge-count

Communities (3)

community

Manifold-aware concept steering in neural representations
members_of
Explores geometry of activation/behavior manifolds to enable selective, non-destructive concept interventions.
Geometric concept representations in neural networks
members_of
Concepts encoded as curved manifolds and circular structures in LLM activation spaces.
Neural geometry as fundamental computational substrate
members_of
Geometric structure in neural representations causally determines computation and behavior across diverse architectures, revealed through analysis of learned manifolds and cyclic concepts.

Claims (1)

claim

Second moments preserve structure that first moments destroy.
associated_with
Core interpretive claim generalizing beyond genomics; argues mean pooling discards information present in covariance.

Related by similarity (8)

cosine ≥ 0.65 · no typed edge

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

Geometry of featuresconcept0.816
Research thread within About Blank concerning the structure and relational properties of neural network feature representations; covariance pooling tangentially supports this thread.
There is a bidirectional relationship between the geometry of representation and behavior across tasks and modalities.claim0.814
Author’s interpretive claim that the shared geometry is general and robust.
The features are often organized in geometrically-related clusters that share a semantic relationship.claim0.804
Decoder cosine similarity maps onto concept similarity.
Features may not be strictly one-dimensional objects; higher-dimensional feature manifolds may exist in model representationshypothesis0.803
Extension of superposition hypothesis to account for continuous families of features
"Representation geometry is a window into the inner world of neural networks."quote0.794
The paper's concluding summary statement asserting the deep interpretive significance of representation geometry.
Features are connected by weights forming circuits, and these circuits can be rigorously studied and understood as meaningful algorithms.claim0.789
Second of three speculative claims asserting that subgraphs of neural networks are tractable and meaningful objects of study
We hypothesize that representation geometry drives model behavior — the geometric structure of internal representations causally shapes what models do externally.hypothesis0.788
The causal hypothesis motivating the use of causality (intervention) as the lens connecting representation and behavior geometry.
How does representation geometry causally drive model behavior?question0.787
The central scientific question the paper addresses through the lens of interventional causality.