claim

active

claim:steering-along-manifolds-provides-better-control-than-linear-steering-when-the-concept-geometry-is-non-linear

Steering along manifolds provides better control than linear steering when the concept geometry is non-linear.

The central thesis of the paper, motivating the shift from linear to geometry-aware manifold steering.

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

Linear steering on Llama-3.1 8B for the days-of-week task cuts across the behavior manifold, producing noisy off-target effects where predicted tokens are not even days of the week.
associated_withsupports
Empirical result demonstrating the failure mode of linear steering when concept geometry is cyclic.
Steering Llama-3.1 8B along the circular representation manifold produces outputs that follow the natural circle of the behavior manifold, cleanly shifting probability mass from Monday through successive days.
associated_withsupports
Core empirical result demonstrating that manifold steering produces on-target, behavior-aligned outputs.

Questions (1)

question

What if the concept being manipulated does not lie on a straight line in the model's representations?
gates
The motivating question that opens the paper and leads to the development of manifold steering.

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-respecting steering produces smooth natural behavioral trajectories while linear steering teleports between non-adjacent concepts.claim0.872
Linear steering produces noisy off-target effects; manifold steering cleanly shifts probability mass between sequential concepts.finding0.871
Core empirical claim comparing steering approaches on cyclic concepts.
manifold steering produces clean probability shifts along natural behavior structure; linear steering cuts across manifold and produces off-target noisy effectsfinding0.839
Empirical demonstration on Llama-3.1-8B that steering along representation manifold aligns outputs with behavior manifold, whereas linear steering does not.
Curved manifolds often represent concepts better than linear directions.claim0.838
Proposes that nonlinear geometric structure is superior to linear feature spaces for capturing semantic content.
Manifold-aware steering is non-trivial IP requiring geometric analysis, not a system-prompt implementation.claim0.833
Linear steering cuts through off-manifold regions and hence produces unnatural outputs.claim0.833
Attribution of failure to Euclidean assumption.
Gap: linear steering assumes Euclidean geometry and does not account for the actual curved geometry of activation manifoldsconcept0.826
The research gap that motivates manifold steering as an alternative to conventional linear approaches
Linear steering is often mismatched with a model's internal representation geometry, producing noisy, off-target effects.claim0.822
The paper's critique of the standard linear steering baseline, supported by the days-of-week demo.