5-token steering pulse

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

• 62% of emotions significantly elevated at 5 tokens after steering pulse endsfinding0.803
  Demonstrates that the majority of emotion features show persistent upregulation shortly after a steering pulse
• All-token steeringmethod0.792
  Baseline steering method that applies intervention at every token generation step, shown to degrade performance at high strengths
• 48 of 171 emotion probes individually significant at token 100 post-steeringfinding0.741
  Shows that causal steering effects persist over long ranges for a substantial fraction of emotion probes
• At 5 tokens after steering pulse ends, 130 of 171 (62%) emotion features are BH-significantly elevated; 14% are suppressed.finding0.733
  Shows immediate causal effect of steering on emotion feature activation
• Do psychological steering results hold beyond 64-token completions?question0.724
  Acknowledged limitation of restricting experiments to 64-token completions
• Steering at 6 layers (strength 0.6 each, total 3.6) outperforms single-layer steering at equivalent total strength for type hint suppressionfinding0.712
  Demonstrates distributed steering is more effective and less accuracy-damaging than concentrated steering.
• Distributing steering strength across multiple layers (6 layers at 0.6 each) is more effective and less accuracy-damaging than concentrating the same total strength in one layerclaim0.694
  Practical finding for optimizing steering setup.
• Feature steering (clamping feature activations)method0.693
  Modifying model behavior by clamping SAE feature activations to specific values during forward pass.