Cross-task and cross-modal validation of manifold steering

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 dynamicsfinding0.792
  Extension of manifold steering validation to video world models and physical dynamics tasks, demonstrating cross-modal generality
• Manifold Steeringconcept0.792
  Central framework: steering neural networks by intervening along the curved manifold where a concept lives, rather than in straight lines through activation space.
• Manifold steering for neural network controlconcept0.785
• Manifold-respecting steering produces smooth natural behavioral trajectories while linear steering teleports between non-adjacent concepts.claim0.784
• Manifold geometry provides a practical blueprint for steering model behavior across diverse tasks and modalities.hypothesis0.777
  The generalizing predictive claim that manifold steering is a broadly applicable framework beyond the days-of-week case study.
• Linear steering produces noisy off-target effects; manifold steering cleanly shifts probability mass between sequential concepts.finding0.774
  Core empirical claim comparing steering approaches on cyclic concepts.
• Steering along manifolds provides better control than linear steering when the concept geometry is non-linear.claim0.770
  The central thesis of the paper, motivating the shift from linear to geometry-aware manifold steering.
• Manifold geometry provides a practical steering blueprint in an image-action model predicting car position on a hill, extending results across modalities.finding0.768
  Cross-modality result from the full paper demonstrating that representation-behavior geometry alignment is not limited to language models.