Term Importance Analysis via Ablation

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

• Zero Ablationmethod0.754
  Intervention type that sets activations to zero, used for interpretability analysis
• Directional Ablationmethod0.749
  Intervention method that removes a direction from residual stream activations to disrupt corresponding behavior
• Feature attribution correlates well with ablation effects, making it an efficient proxy for causal effect.claim0.744
  Gradient-based attribution approximates ablation impact, enabling fast search for causally important features.
• Grafting and Ablationmethod0.742
  Classical techniques to interrogate regulative capacity of embryos and neural crest by tissue removal or transplantation.
• Surgical Ablation Propertyconcept0.736
  Property requiring that ablating a truth direction shifts model output from truthful to false without other side effects
• Causal importanceconcept0.730
  A measure of whether a subcomponent is necessary to reproduce model behavior on a specific prompt, predicted by the causal importance network.
• Feature ablation (zeroing feature activations)method0.727
  Clamping a feature's value to zero to measure its causal effect on model output.
• Importance Scoringmethod0.724
  Weighted Spearman correlation that corrects for sampling bias in automated interpretability evaluation