Why does L2 regularisation increase probe causal efficacy (selectivity)?

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

• Are high-accuracy probe representations also causally relevant for the task?question0.749
  Question raised by the discrepancy between DAS IIA and linear probe accuracy in Case Study II
• Probe-based data attribution effectively reduces harmful behaviors via data interventionsclaim0.740
  Authors' central interpretive assertion that their method meaningfully mitigates unwanted behaviors.
• A probe may achieve high performance even on representations that are not causally relevant for the taskclaim0.738
  Key interpretive claim from Case Study II distinguishing probe accuracy from causal relevance
• Feature attribution correlates well with ablation effects, making it an efficient proxy for causal effect.claim0.737
  Gradient-based attribution approximates ablation impact, enabling fast search for causally important features.
• Mass-mean probe directions outperform LR and CCS in causal intervention experiments (NIE) in 7/8 experimental conditionsfinding0.737
  Core result showing MM is superior to LR for causal implication despite similar classification accuracy
• The target vs. off-target probe area metric quantifies steering selectivity and distinguishes selectively steerable from entangled interventions.claim0.736
  Justification for the novel metric introduced in the paper
• Direct probes over learned activations in standard basis may fail to reveal the actual causal role of representations because they are highly distributedclaim0.736
  Supported by the finding that non-trivial rotations are required to find aligned representations.
• Why were interventions with mass-mean probe directions extracted from the likely dataset so effective, despite these probes not being accurate at classifying true/false statements?question0.733
  Open question raised in §7.1 about an unexplained anomalous result