hypothesis

active

hypothesis:manifold-geometry-provides-a-practical-blueprint-for-steering-model-behavior-across-diverse-tasks-and-modalities

Manifold geometry provides a practical blueprint for steering model behavior across diverse tasks and modalities.

The generalizing predictive claim that manifold steering is a broadly applicable framework beyond the days-of-week case study.

Source paper

extracted_from

Steering Along Manifolds to Control Neural Networks

Neighborhood — ranked by edge-count

Papers (1)

paper

Steering Along Manifolds to Control Neural Networks
introduces

Findings (2)

finding

Analogous alignment between representation manifold and behavior manifold is found across months, letters, ages, and synthetic in-context learning tasks in language models.
associated_withsupports
Generalization finding from the full paper extending beyond days-of-week to other structured concepts.
Manifold geometry provides a practical steering blueprint in an image-action model predicting car position on a hill, extending results across modalities.
associated_with
Cross-modality result from the full paper demonstrating that representation-behavior geometry alignment is not limited to language models.

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.

Manifold steering demonstrates bidirectional geometry-behavior link in a video world model on tasks with geometry corresponding to physical dynamicsfinding0.844
Extension of manifold steering validation to video world models and physical dynamics tasks, demonstrating cross-modal generality
There is a clear bidirectional relationship between the geometry of behavior and representation: steering along representation manifolds follows behavior manifolds, and vice versa.claim0.825
The paper's finding that the alignment holds in both directions — from representation to behavior and from behavior back to representation space.
manifold geometry principles extend to months, letters, ages, and in-context learning tasks across modalitiesfinding0.823
Evidence that the weekday cyclic structure is not anomalous but reflects broader principle of concept geometry.
Steering along manifolds provides better control than linear steering when the concept geometry is non-linear.claim0.814
The central thesis of the paper, motivating the shift from linear to geometry-aware manifold steering.
Manifold-aware steering is non-trivial IP requiring geometric analysis, not a system-prompt implementation.claim0.810
Manifold-respecting steering produces smooth natural behavioral trajectories while linear steering teleports between non-adjacent concepts.claim0.810
Curved manifolds often represent concepts better than linear directions.claim0.806
Proposes that nonlinear geometric structure is superior to linear feature spaces for capturing semantic content.
manifold steering produces clean probability shifts along natural behavior structure; linear steering cuts across manifold and produces off-target noisy effectsfinding0.805
Empirical demonstration on Llama-3.1-8B that steering along representation manifold aligns outputs with behavior manifold, whereas linear steering does not.