Probes trained on h_b activations achieve perfect test accuracy in every case; h_s probes achieve perfect accuracy in only 0.60% of cases

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

• For LLaMA-2-70B, probes trained on larger_than+smaller_than achieve >95% accuracy on sp_en_trans regardless of probing techniquefinding0.800
  Striking cross-domain generalization result supporting the claim that larger models represent abstract truth
• Probes trained on A1 degrade significantly when evaluated on A2 and more on A3; training on A2 achieves only AUROC ~0.65 on A3.finding0.792
  Shows rapid generalization decay for arithmetic truth directions with each additional operation.
• Probes trained under different explicit instruction prompts (ask-correct, ask-t/f, ask-able, ask-arith) are highly aligned with each other in cosine similarity.finding0.790
  Shows the passive vs. active divide is more important than the specific wording of instructions.
• Probes trained on the likely dataset perform worse than chance on datasets with anti-correlations between text probability and truthfinding0.789
  Shows that truth representations are not reducible to text probability representations
• A probe may achieve high performance even on representations that are not causally relevant for the taskclaim0.784
  Key interpretive claim from Case Study II distinguishing probe accuracy from causal relevance
• Probes trained under different explicit instruction variants are highly aligned with each other despite different wording.claim0.782
  Shows the key divide is passive vs. active framing, not the specific wording of instructions.
• MM probes trained on larger_than+smaller_than achieve lower NIE than those trained on cities+neg_cities despite higher classification accuracy on sp_en_transfinding0.780
  Dissociation between classification accuracy and causal implication; training on opposites does not always help causally
• MM probe trained on likely dataset achieves NIE of 0.70 (false→true) on LLaMA-2-13B, surprisingly strong but weaker than truth probesfinding0.775
  Likely-trained MM probe is a surprisingly effective causal baseline due to correlation between truth and probability on sp_en_trans